WO2022266266A1

WO2022266266A1 - Methods and compositions for evaluating and treating blood glucose dysregulation

Info

Publication number: WO2022266266A1
Application number: PCT/US2022/033693
Authority: WO
Inventors: Nan Shen; Nevenka Dimitrova; Han-Chen Cleo HO; Ying Cai; Momchilo VUYISICH; Damon TANTON; Guruduth S. Banavar
Original assignee: Viome Life Sciences Inc
Current assignee: Viome Life Sciences Inc
Priority date: 2021-06-15
Filing date: 2022-06-15
Publication date: 2022-12-22
Anticipated expiration: 2023-12-15

Abstract

Provided herein are methods for determining risk of developing or progressing glucose dysregulation in a subject, determining likelihood of responding to metformin in a subject suffering from glucose dysregulation, and administering a treatment to a subject based on the determination.

Description

METHODS AND COMPOSITIONS FOR EVALUATING AND TREATING BLOOD

GLUCOSE DYSREGULATION

[0001] This application claims the benefit of U.S. Provisional Application No. 63/210,906, filed June 15, 2021, which application is incorporated herein by reference.

BACKGROUND

[0002] Type 2 diabetes (T2D) is the most prevalent endocrine disease, affecting more than 400 million people worldwide, and this number is expected to rise to 700 million by the year 2025 (Zhou et al., 2016) (Chatterjee et al., 2017). The increased prevalence of T2D is a major challenge for healthcare systems globally and novel tools are needed to reduce the burden of T2D. Recognizing and treating the early stages (and even predisease) is the most cost effective way to treat T2D, before end-organ damage, such as heart disease, stroke, renal disease, blindness and limb amputation, ensues. (Bailes, 2002). Multiple genomic and transcriptomic studies have been conducted on the analysis of human host samples and animal models to elucidate the molecular mechanisms of disease progression (Lawlor et al., 2017; Segerstolpe et al., 2016; Sengupta et al., 2009). Although it is a complex disorder influenced by both genetic and environmental components, central obesity (visceral fat) is known to be the driving risk factor (Hu, 2011).

[0003] In recent years, the gut microbiome has emerged as a prominent new frontier in medical research, having been found to be implicated in a range of chronic and acute health conditions and outcomes. Gut microbiome research has primarily focused on genomic DNA. Multiple studies have shown that the gut microbiome is altered in individuals with metabolic disorders, such as obesity (Tumbaugh et al., 2009) (Ley et al., 2006), T2D (Larsen et al., 2010) (Gurung et al., 2020) and progression of glucose intolerance (Zhang et al., 2013), and that the gut microbiome may be a causal factor in the development of these disease processes. (Wang et al., 2020). (Ridaura et al., 2013). Additionally, research has evaluated the predictive power of the gut microbiome for early T2D risk detection (Li et al., 2020). Findings in the existing metagenomic studies, indicated that Ruminococcus, Fusobacterium, Blautia are positively associated, while Bifidobacterium, Bacteroides, Faecalibacterium, Akkermansia and Roseburia are negatively associated with type 2 diabetes (Gurung et al., 2020).

[0004] While previous metagenomic studies have revealed compelling insights into the role that microorganisms play in the development of T2D, these approaches have described the presence of organisms and genes, rather than activity of these microbiome elements. Using non- invasive fecal samples, clinical grade) and fully automated metatranscriptomic technology (Hatch et al., 2019) has the potential to facilitate large-scale public health applications, including early diagnosis and risk assessment for T2D.

INCORPORATION BY REFERENCE

[0005] All publications, patents, and patent applications mentioned in this specification are herein incorporated by reference to the same extent as if each individual publication, patent, or patent application was specifically and individually indicated to be incorporated by reference.

BRIEF DESCRIPTION OF THE DRAWINGS [0006] The novel features of the invention are set forth with particularity in the appended claims. A better understanding of the features and advantages of the present invention will be obtained by reference to the following detailed description that sets forth illustrative embodiments, in which the principles of the invention are utilized, and the accompanying drawings of which:

[0007] Figure 1. Discovery cohort, used for statistical analysis and model training/development. (Note: Metformin (n=80) are people using only metformin; other drugs (n=23) are people using drugs other than metformin; the remainder of medicated (n=150) were people using metformin along with other drugs, and are not shown here, nor were used in any analysis.)

[0008] Figure 2: Progression of disease: differentially expressed features in a color-bar scatter plot (KW statistics, FDR p-value < 0.05). Panels present scatter plots of resulting patient features which include (a) differentially expressed species shown as circles between non- medicated T2D vs non-medicated non-T2D and (b) differentially expressed KOs shown as triangles between non-medicated T2D vs non-medicated non-T2D. Filled markers represent obesity related features, and hollow markers represent T2D-unique features.

[0009] Figure 3: a) Differential prevalence of species: left panel represents the differential prevalence between pre-T2D vs non-T2D, and the right panel represents the differential prevalence between T2D vs non-T2D b) Differential expression of species between pre-T2D vs non-T2D c) Differential prevalence of KOs, grouped by KEGG category: left panel represents the differential prevalence between non-medicated pre-T2D vs non-medicated non-T2D, and the right side represents the differential prevalence between non-medicated non T2D vs -medicated d) Differential expression of KOs, grouped by KEGG category: left panel represents the differential expression between non-medicated pre-T2D vs non-medicated non-T2D, and the right side represents the differential expression between non-medicated T2D vs non-medicated non-T2D . [0010] Figure 4 Important features in T2D disease progression classification model distinguishing T2D patients and prediabetic patients from presumed healthy using SVM (species with differential expression > 0.1; top 300 KOs with the highest coefficients and differential expression > 0.2) a) Important species in T2D classifiers: left side shows non-medicated pre-T2D vs non-T2D important species; right side shows non-medicated T2D vs non-T2D important species Important KOs in classifiers grouped by KEGG L3 category b) Important KOs in classifiers grouped by KEEG L3 category. Left side shows non-medicated pre-T2D vs non-T2D important KOs, blue triangles show higher expression in the non-medicated T2D group, red triangles show higher expression in pre-diabetic group; right side shows non-medicated T2D vs non-T2D important KOs, blue triangles show higher expression in the non-diabetic group, red triangles show higher expression in T2D group.

[0011] Figure 5 Bar plots and probability density distribution plots of Ruminococcus and Blautia features (a) Prevalence bar plot (top) and expression violin plots (bottom) of the species in the Ruminococcus genus (b) Prevalence bar plot (top) and expression violin plots (bottom) of the species in the Blautia genus

[0012] Figure 6 Progression of disease: non-diabetic, pre-diabetic and T2D diversity and richness a) Taxa and KO richness; b) Taxa and KO Shannon diversity [0013] Figure 7 Statistically significant richness, diversity (a-b) and feature scatter plots distinguishing T2D vs. general population using descriptive statistics (species FDR p value < 0.05; KOs FDR p value < 0.00001) (c-1). Venn diagrams representing overlapping features shared between the T2D progression and the diagnostic statistical analysis (g-j) (a) taxa and KO richness; (b) taxa and KO Shannon diversity; (c) differential prevalence of species; (d) differential expression of species; (e) differential prevalence of KOs; (1) differential expression of KOs; (g) sharing significantly prevalent species between progression and diagnostic descriptive analysis; (h) sharing significantly prevalent KOs between progression and diagnostic descriptive analysis; (i) sharing significantly differentially expressed species between progression and diagnostic descriptive analysis; (j) sharing significantly differentially expressed KOs between progression and diagnostic descriptive analysis.

[0014] Figure 8 Type 2 diabetes diagnostic analysis and support vector machine approach a) ROC curves of diagnostic classifiers b) Risk scores of the validation cohort.

[0015] Figure 9 Important diagnostic model features using support vector machine approach (species with differential expression > 0.1; top 300 KOs with the highest coefficients and differential expression > 0.2) (a-b). c) correlation between T2D risk score and BMI in the validation cohort (n=2406) (a) Important species in T2D vs. non-diabetic classifier; (b) Important KOs in T2D vs. non-diabetic classifier; (c) Correlation between BMI and T2D risk score (rho=0.16, p=5.29e-19

[0016] Figure 10 T2D metformin treated patients with good control of HbAlc (<6.5) (n=22) vs T2D metformin treated patients who cannot control their HbAlc (>=6.5) (n=41) AUC = 0.74+/-0.11 (a) ROC curves of metformin treatment classifiers (b) Summary of important features and potential functions of the metformin models (c) Important differentially expressed species in T2D non-medicated patients who are able to control HbAlc (<=6.5) in red vs. patients who are non-medicated and not able to control HbAlc (>6.5) in blue (d) Differentially expressed KOs in T2D non-medicated patients who are able to control HbAlc (<=6.5) in red vs. patients who are non-medicated and not able to control HbAlc (>6.5) in blue. Here are visualized species with differential expression > 0.1; top 300 KOs with the highest importance coefficients and differential expression > 0.2

[0017] Figure 11 Richness and diversity in treatment models (a-b) T2D metformin treated patients with good control of HbAlc (<6.5) (n=22) vs T2D metformin treated patients who can not control their HbAlc (>=6.5) (n=41); (c-d) T2D non-medicated responders with HbAlc <6.5 (n=29) vs T2D non-medicated non-responders HbAlc >=6.5 (n=30). (a) Taxa and KO richness in T2D metformin responder v.s. non-responder; (b) Taxa and KO Shannon-diversity in T2D metformin responder v.s. non-responder; (c) Taxa and KO richness in T2D non-medicated responder v.s. non-responder; (d) Taxa and KO Shannon-diversity in T2D non-medicated responder v.s. non-responder

[0018] Figure 12 Biological interpretation of the type 2 diabetes metatranscriptomic analysis and derived features. Orange and green circles represent features enriched in the diseased (preT2D or T2D) and non-diabetic cohorts, respectively.

DETAILED DESCRIPTION OF THE INVENTION [0019] Provided herein are methods and compositions for predicting and treating glucose dysregulation, such as Type 2 diabetes (T2D), pre-diabetes, or even stages prior to pre-diabetes, based at least in part on gut metatranscriptomic information. In certain embodiments, likely control or no control of blood glucose by metformin is determined and/or modified based on metatranscriptomic evidence. While previous metagenomic studies have revealed compelling insights into the role that microorganisms play in the development of T2D, these approaches have described the presence of organisms and genes, rather than activity of these microbiome elements, as is used here in a metatranscriptomic analysis. Using non-invasive fecal samples, and fully automated metatranscriptomic technology (Hatch et ak, 2019) has the potential to facilitate large-scale public health applications, including early diagnosis and risk assessment for T2D. [0020] The current standard of care for diagnosing and treating Type 2 diabetes is based largely on HbAlc values, with aHbAlc of less than 5.7% considered non-diabetic, 5.7-6.4% considered prediabetic, and 6.5% or higher diabetic. Prediabetic and diabetic individuals are often treated with lifestyle modification, medication, or both. Metformin is the most common diabetes medication used. Provided herein are methods for determining whether or not an individual who, e.g., is considered non-diabetic by conventional methods, for example as determined by HbAlc, may nonetheless have elevated risk for progressing to prediabetes or diabetes, based at least in part on metatranscriptomic information from the individual. In certain embodiments, the individual is prediabetic and the assessment is risk for diabetes. A risk score for the individual for progressing to prediabetes or diabetes can be based on an algorithm that includes metatranscriptomic data. Such data may show that an individual who would be considered non-diabetic by conventional methods, e.g., an individual with aHbAlc value of less than 5.7%, may be at elevated risk for development of prediabetes and/or diabetes, and treatment may be initiated to slow or prevent development of prediabetes and/or diabetes in the individual. Thus, treatment, which can include lifestyle modification and/or medication, can be initiated and/or monitored based on the information. Also provided herein are methods for determining whether or not an individual who is prediabetic, for example as determined by HbAlc, may have elevated risk for progressing to diabetes, based at least in part on metatranscriptomic information from the individual. A risk score for the individual for progressing to diabetes can be based on an algorithm that includes metatranscriptomic data. Treatment, which can include lifestyle modification and/or medication, can be initiated and/or monitored based on the information. Further provided herein are methods for determining whether or not an individual will experience control of blood glucose with treatment with metformin based at least in part on metatranscriptomic information from the individual; in certain embodiments, treatment initiation or modulation may be based on such information.

[0021] Methods and compositions are based on collection of information from one or more samples from an individual that allow evaluation of gut microbiome; typically such samples will be stool samples. Sample collection, ambient temperature sample preservation, total RNA extraction, physical removal of ribosomal RNAs (rRNAs), preparation of directional libraries, e.g., Illumina libraries, and sequencing, e.g., Illumina sequencing, may be performed by any suitable method, for example, as describe in Hatch et al. (2019), A robust metatranscriptomic technology for population-scale studies of diet, gut microbiome, and human health. Inti. J. Genom. https://doi.org/10.1155/2019/1718741, and in Patent Publication Nos. WO2019209753; US20190153438; WO2020247983; W02020076874; W02020051559; WO2019113563; W02020168015; , and Example 1. An RNA stabilizer can be added to the sample to preserve RNA during sampling and shipping. Phenotypic data may also be collected from the individual, and can include any data relevant to assessing risk of diabetes and/or to microbiome modification, such as age, height, weight, BMI (can be calculated from weight and height), gender, ethnicity, use of medications, especially those that may affect the microbiome, such as antibiotics, proton pump inhibitors or acid suppressants, health history, which can include, e.g., abdominal surgery and/or diseases that may affect gut microbiome, such as IBS, IBD, colon cancer; HbAlc values, blood glucose values, blood insulin values, glucose tolerance test values (blood glucose and, optionally, blood insulin), and other suitable characteristics as will be apparent to one of skill in the art. In certain embodiments, phenotypic data is also used in the evaluation of risk of glucose dysregulation.

[0022] In certain embodiments, provided is a method of determining risk for developing diabetes or prediabetes in an individual, for example an individual who is nondiabetic by conventional measure, e.g., an HbAlc of less than 5.7%, , or determining risk of developing or progressing blood glucose dysregulation, by evaluating mRNA in a stool sample from the individual, and, in certain embodiments, treating the individual based on their level of risk, for example, through administration of medication, dietary changes, supplements, such as pre-biotic or pro-biotic supplements, or other lifestyle changes (e.g., exercise). Dosage of each of the treatments can be based on features of the risk assessment. In certain embodiments, provided is a method of determining risk for developing diabetes in an individual, for example an individual who is prediabetic by conventional measure, e.g., an HbAlc of 5.7 to 6.4%, or determining risk of developing or progressing blood glucose dysregulation, by evaluating mRNA in a stool sample from the individual, and, in certain embodiments, treating the individual based on their level of risk, for example, through administration of medication, dietary changes, supplements, such as pre-biotic or pro-biotic supplements, or other lifestyle changes (e.g., exercise). Dosage of each of the treatments can be based on features of the risk assessment.

[0023] Thus, provided herein is a method of determination of risk for developing diabetes or prediabetes, or determining risk of developing or progressing blood glucose dysregulation, involves providing a biological sample from the individual that comprises a gut microbiome, for example a stool sample; sequencing nucleic acids from the sample to provide sequence information, where the nucleic acids comprise RNA, e.g., mRNA; determining from the sequence information one or more features of the nucleic acid sequences; and executing a classification model, for example, by computer, that infers, from the one or more feature of in the feature set, a measure of risk for development of a prediabetic state, a diabetic state, or development or progression of glycemic dysregulation in the individual. The sample can be analyzed for, e.g., transcriptional information, such as taxonomic information and KO classification. Thus in certain embodiments, the one or more features comprise (1) determinations of KEGG-Orthologs (KOs) for gene-level activity within the sample the determinations of KOs comprise at least 5, 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 85, 90, 95, 100, 110, 120, 130, 150, 170, or 200 different KOs and/or not more than 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 85, 90, 95, 100, 110, 120, 130, 150, 170, 200 or 300 KOs. Additionally or alternatively, the one or more features comprise (2) measures of active microbial species within the sample, wherein the one or more measures are included in a feature set. The measures of active microbial species can include measures of species from at least 1, 2,

5, 7, 10, 12, 15, 17, 20, 22, 25, 30, 40, or 50 genuses and/or not more than 2, 5, 7, 10, 12, 15, 17, 20, 22, 25, 30, 40, 50 or 70 genuses. In certain embodiments the biological sample comprises a stool sample. In certain embodiments the individual is a human. In certain embodiments the individual has an HbAlc of less than 5.7% and is conventionally classified as nondiabetic. In certain of these embodiments the measure of risk for the individual indicates that the individual is at risk of developing glycemic dysregulation. In certain embodiments the individual has an HbAlc of 5.7-6.4% and is conventionally classified as prediabetic. In certain of these embodiments the measure of risk for the individual indicates that the individual is at risk of progressing glycemic dysregulation. In certain embodiments the individual has an HbAlc of 6.5% or greater and is conventionally classified as diabetic. In certain of these embodiments the measure of risk for the individual indicates that the individual is at risk of progressing glycemic dysregulation. In certain embodiments, the sequence information comprises metatranscriptomic information. In certain embodiments the feature set used by the classification algorithm includes at least measures of activity of one or more microbial taxa. In certain embodiments the feature set used by the classification algorithm includes at least measures of taxa involved in one or more metabolic pathways. In certaine embodiments the classification model uses expression levels for one or more species of one or more genus shown in Figure 9a. In certain embodiments the classification model uses expression levels for species of at least, exactly, and/or no more than any of 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30 of the genuses shown in Figure 9a. In certain embodiments, a higher risk of developing or progressing glycemic dysregulation is determined, at least in part, from relatively lower expression of Ruminococus species, e.g., as compared to expression levels found in nondiabetic individuals. In certain embodiments the Ruminococus species comprise Ruminococcus bicirculans, R. callidus, and/or R. champallensis . In certain embodiments, a higher risk of developing or progressing glycemic dysregulation is determined, at least in part, from relatively lower expression of Blautia species in the sample, e.g., as compared to expression levels found in nondiabetic individuals. In certain embodiments the Blautia species comprise Blautia massiliensis and/or Blautia Mar seille-P 3087. In certain embodiments the classification model uses at least, exactly, and/or no more than any of 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14,

15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40,

41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66,

67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92,

93, 94, or 95 of the KOs selected from those of Figure 9b. In certain embodiments the method further comprises outputting the inference to a user interface device or to a computer-readable memory.

[0024] In certain embodiments an individual is found to be at risk for developing or progressing glycemic dysregulation, for example by the method of the above paragraph, and a method comprises delivering and/or administering to the individual one or more therapeutic interventions to decrease glycemic dysregulation or the development of glycemic dysregulation. The choice of therapy, including dosing regimen, can be based at least in part on the determination of risk and/or on metatranscriptomic analysis. In certain embodiments the therapeutic intervention comprises administering one or more pharmaceutical agents to the individual. In certain embodiments the one or more pharmaceutical agents comprise metformin, sodium glucose co-tranporter 2 inhibitor (SGLT2i), a glucagon-like peptide 1 receptor agonist (GLP-RA), or a combination thereof. In certain embodiments the pharmaceutical agent comprises metformin. In certain embodiments 500 mg of metformin is taken daily. In certain embodiments 1000 mg of metformin is taken daily. In certain embodiments 1500 mg of metformin is taken daily. In certain embodiments 2000 mg of metformin is taken daily. In certain embodiments 2500 mg of metformin is taken daily. In certain embodiments, one or more therapeutic interventions is used to modulate the gut microbiome to improve responsiveness to metformin. In certain embodiments one or more pro-biotic or pre-biotic supplements, or a combination of both, is given to improve response to metformin. In certain embodiments, one or more dietary modifications, such as one or more of those modifications described elsewhere herein, is given to improve response to metformin. In certain embodiments the pharmaceutical agent comprises a SGLT2i selected from the group consisting of canagliflozin, dapagliflozin, and empagliflozin. In certain embodiments the pharmaceutical agent comprises a GLP-RA selected from the group consisting of Dulaglutide, liraglutide, lixisenatide, Exenatide, Exenatide extended release, Semaglutide, injection, and Semaglutide, oral. In certain embodiments the therapy comprises a dietary therapy. In certain embodiments the dietary therapy comprises eating a whole foods plant-based diet and reducing or eliminating meat; reducing or eliminating cheese consumption, eating the largest meal of the day at midday, or a combination thereof. In certain embodiments the dietary therapy comprises eating a diet of an average of less than 200, 150, 100, or 50 grams of carbohydrate daily, periodically fasting for at least 16 hours, or a combination thereof. In certain embodiments the therapy comprises exercise, for example resistance training, endurance training, high intensity training, or a combination thereof. In certain embodiments therapy comprises administration of one or more supplements. In certain embodiments the supplements comprise one or more pre-biotics, one or more probiotics, or a combination thereof. In certain embodiments the one or more prebiotics and/or probiotics are chosen to modulate the gut microbiome of the individual toward a biome that favors decreased glucose dysregulation. [0025] Provided herein is a method for determining whether or an individual is able to control blood glucose with metformin comprising a) providing a biological sample from the individual; b) sequencing nucleic acids from the sample to provide sequence information, wherein the nucleic acids comprise RNA, e.g., mRNA; c) determining, from the sequence information one or more features of the nucleic acid sequences; d) executing by computer a classification model that infers, from one or more features in the feature set, a measure of likelihood of control of blood glucose with metformin for the individual. Control of blood glucose may be determined in any suitable manner; in certain embodiments, control of blood glucose is defined as reducing HbAlc to 6.5% or lower. In certain embodiments the one or more features comprise (1) determinations of KEGG-Orthologs (KOs) for gene-level activity within the sample. In certain embodiments the determinations of KOs comprise at least 5, 10, 15, 20,

25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 85, 90, 95, 100, 110, 120, 130, 150, 170, or 200 different KOs and/or not more than 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 85, 90, 95, 100, 110, 120, 130, 150, 170, 200 or 300 KOs. Alternatively or additionally, the one or more features comprise (2) measures of active microbial species within the sample, wherein the one or more measures are included in a feature set. In certain embodiments measures of active microbial species include measures of species from at least 1, 2, 5, 7, 10, 12, 15, 17, 20, 22, 25, 30, 40, or 50 genuses and/or not more than 2, 5, 7, 10, 12, 15, 17, 20, 22, 25, 30, 40, 50 or 70 genuses. In certain embodiments the biological sample comprises a stool sample. In certain embodiments the individual is a human. In certain embodiments the individual has an HbAlc of less than 5.7% and is conventionally classified as nondiabetic. In certain embodiments the individual has an HbAlc of 5.7-6.4% and is conventionally classified as prediabetic. In certain embodiments the individual has an HbAlc of 6.5% or greater and is conventionally classified as diabetic. In certain embodiments the sequence information comprises metatranscriptomic information. In certain embodiments the feature set used by the classification algorithm includes at least measures of activity of one or more microbial taxa. In certain embodiements the feature set used by the classification algorithm includes at least measures of taxa involved in one or more metabolic pathways. In certain embodiments the classification model uses expression levels for one or more species of one or more genuses shown in Figure 10c. In certain embodiments the classification model uses expression levels for species of at least, exactly, and/or no more than any of 1, 2, 3, 4, or 5, of the genuses shown in Figure 10c. In certain embodiments the classification model uses at least, exactly, and/or no more than any of 1, 2, 3, 4, 5, 6, 7, 8, 9, 10,

11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62

63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, or 81, of the KOs selected from those of Figure lOd.

[0026] In certain embodiments an individual is treated with metformin based on the determination of the above paragraph. In certain embodiments 500 mg of metformin is taken daily. In certain embodiments 1000 mg of metformin is taken daily. In certain embodiments 1500 mg of metformin is taken daily. In certain embodiments 2000 mg of metformin is taken daily. In certain embodiments 2500 mg of metformin is taken. In certain embodiments, one or more therapeutic interventions is used to modulate the gut microbiome to improve responsiveness to metformin. In certain embodiments one or more pro-biotic or pre-biotic supplements, or a combination of both, is given to improve response to metformin. In certain embodiments, one or more dietary modifications, such as one or more of those modifications described elsewhere herein, is given to improve response to metformin.

[0027] Also provided herein is a method comprising: a) providing biological samples from each of a first set of subjects, a second set of subjects, and a third set of subjects, wherein the biological samples comprise a stool samples, and wherein the first set of subjects have been diagnosed with Type 2 diabetes, the second set of subjects have been diagnosed as prediabetic, and the third set of subjects have been diagnosed as nondiabetic; b) sequencing nucleic acids in the biological samples to provide sequence information; and c) performing a statistical analysis on the sequence information to produce a model that predicts a risk of developing or progressing glucose dysregulation in an individual. In certain embodiments the statistical analysis comprise a model developed by machine learning. In certain embodiments the statistical analysis comprises an analysis selected from correlational, Pearson correlation, Spearman correlation, chi-square, comparison of means (e.g., paired T-test, independent T-test, ANOVA) regression analysis (e.g., simple regression, multiple regression, linear regression, non-linear regression, logistic regression, polynomial regression stepwise regression, ridge regression, lasso regression, elasticnet regression) and non-parametric analysis (e.g., Wilcoxon rank-sum test, Wilcoxon sign- rank test, sign test).

[0028] Also provided is a system comprising: (i) a computer comprising: (a) a processor; and (b) a memory storage module, coupled to the processor, the memory storing a module comprising: (1) nucleic acid sequence information from a biological sample from a subject comprising a gut microbiome; (2) a classification model which, based on values including the measurements, classifies the subject as at risk for developing or progressing glucose dysregulation, wherein the classification model is selected to have a sensitivity of at least 75%, at least 85% or at least 95%; and (3) computer executable instructions for implementing the classification model on the test data.

[0029] Also provided herein is a method for developing a computer model for inferring, from feature data, a risk of developing or progressing glucose dysregulation in a subject comprising: a) training a machine learning algorithm on a training data set, wherein the training data set comprises, for each of a plurality of subjects, (1) a class label classifying a subject as nondiabetic, prediabetic, or diabetic; and (2) feature data comprising quantitative measures for each of a plurality of features selected from gut microbiome transcriptome expression, and wherein the machine learning algorithm develops a model that infers a class label for a subject based on the feature data.

[0030] Also provided herein is a method that infers a risk of developing or progressing glucose dysregulation in a subject, the method comprising: (a) providing a data set comprising, for the subject, feature data for each of a plurality of features selected from gut microbiome transcriptome gene expression data and taxa activity data; and (b) executing a computer model on the data set to infer the risk of developing or progressing glucose dysregulation in the subject. [0031] Also provided herein is a software product comprising a computer readable medium in tangible form comprising machine executable code, which, when executed by a computer processor, infers a risk of developing or progressing glucose dysregulation in a subject by: (a) accessing a data set comprising, for a subject, feature data for each of a plurality of features selected from gut microbiome transcriptome gene expression data and taxa activity data; and (b) executing a computer model on the data set to infer the risk of developing or progressing glucose dysregulation in the subject.

[0032] Also provided herein is a method of providing therapy to decrease a risk of developing or progressing glucose dysregulation or to normalize glucose regulation in a subject comprising: (a) inferring the presence of a risk of developing or progressing glucose dysregulation in a subject according to a method as described herein; and (b) administering a therapeutic intervention to the subject effective to decrease the risk of developing or progressing glucose dysregulation or to normalize glucose regulation in the subject.

[0033] Also provided herein is a method for diagnosing and treating developing or progressing glucose dysregulation in a subject, the method comprising: (a) receiving from a subject a sample comprising a gut microbiome; (b) determining nucleic acid sequences of a microorganism component of the sample; (c) determining alignments of the nucleic acid sequence to reference nucleic acid sequences associated with developing or progressing glucose dysregulation; (d) generating a microbiome feature dataset for the subject based upon the alignments; (e) generating an inference of the risk of developing or progressing glucose dysregulation in the subject upon processing the microbiome feature dataset with an inference model derived from a population of subjects; and (1) at an output device associated with the subject, providing a therapy to the subject with the risk of developing or progressing glucose dysregulation upon processing the inference with a therapy model designed to treat the glucose dysregulation. Also provided herein is a method comprising: (a) measuring, in a sample from a subject comprising an gut microbiome, activity of one or more biomarkers selected from Figure 9; (b) inferring, from the measurements, a risk of developing or progressing glucose dysregulation in the subject; and (c) delivering to the subject a therapeutic intervention effective to decrease the risk of developing or progressing glucose dysregulation or to normalize glucose regulation in the subject.

[0034] Also provided herein is a method comprising: a) providing biological samples from each of a first set of subjects and a second set of subjects having glucose dysregulation and having been subject to a therapeutic intervention, wherein the biological samples comprise a gut microbiome and wherein the first set of subjects responded positively to the therapeutic intervention and the second set of subjects did not respond positively to the therapeutic intervention; b) sequencing nucleic acids in the biological samples to provide sequence information; and c) performing a statistical analysis on the sequence information to produce a model that infers having a positive response or lack of positive response to the therapeutic intervention in a subject.

[0035] Also provided herein is a method of treating a subject with glucose dysregulation comprising: (a) inferring that the subject will respond positively to each of one or more therapeutic interventions by executing a model on nucleic acid information from a biological sample from the subject comprising a gut microbiom; and (b) administering to the subject one or more therapeutic interventions to treat the glucose dysregulation; In certain embodiments the therapeutic intervention comprises administration of metformin.

EMBODIMENTS

[0036] In embodiment 1 provided is a method for determining whether an individual is at risk for development or progression of glycemic dysregulation comprising a) providing a biological sample from the individual; b) sequencing nucleic acids from the sample to provide sequence information, wherein the nucleic acids comprise mRNA; c) determining, from the sequence information one or more features of the nucleic acid sequences; d) executing by computer a classification model that infers, from one or more features in the feature set, a measure of risk for development or progression of glycemic dysregulation in the individual. In embodiment 2 provided is the method of embodiment 1 wherein the one or more features comprise (1) determinations of KEGG-Orthologs (KOs) for gene-level activity within the sample. In embodiment 3 provided is the method of embodiment 1 or 2 wherein the determinations of KOs comprise at least 5, 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 85, 90, 95, 100, 110, 120, 130, 150, 170, or 200 different KOs and/or not more than 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 85, 90, 95, 100, 110, 120, 130, 150, 170, 200 or 300 KOs. In embodiment 4 provided is the method of any preceding embodiment wherein the one or more features comprise (2) measures of active microbial species within the sample, wherein the one or more measures are included in a feature set. In embodiment 5 provided is the method of any preceding embodiment wherein the measures of active microbial species include measures of species from at least 1, 2, 5, 7, 10, 12, 15, 17, 20, 22, 25, 30, 40, or 50 genuses and/or not more than 2, 5, 7, 10, 12, 15, 17, 20, 22, 25, 30, 40, 50 or 70 genuses. In embodiment 6 provided is the method of any preceding embodiment wherein the biological sample comprises a stool sample. In embodiment 7 provided is the method of any preceding embodiment wherein the individual is a human. In embodiment 8 provided is the method of any preceding embodiment wherein the individual has an HbAlc of less than 5.7% and is conventionally classified as nondiabetic. In embodiment 9 provided is the method of any preceding embodiment wherein the measure of risk for the individual indicates that the individual is at risk of developing glycemic dysregulation. In embodiment 10 provided is the method of any preceding embodiment wherein the individual has an HbAlc of 5.7-6.4% and is conventionally classified as prediabetic. In embodiment 11 provided is the method of any preceding embodiment wherein the measure of risk for the individual indicates that the individual is at risk of progressing glycemic dysregulation. In embodiment 12 provided is the method of any preceding embodiment wherein the individual has an HbAlc of 6.5% or greater and is conventionally classified as diabetic. In embodiment 13 provided is the method of any preceding embodiment wherein the measure of risk for the individual indicates that the individual is at risk of progressing glycemic dysregulation. In embodiment 14 provided is the method of any preceding embodiment wherein the sequence information comprises metatranscriptomic information. In embodiment 15 provided is the method of any preceding embodiment wherein the feature set used by the classification algorithm includes at least measures of activity of one or more microbial taxa. In embodiment 16 provided is the method of any preceding embodiment wherein the feature set used by the classification algorithm includes at least measures of taxa involved in one or more metabolic pathways. In embodiment 17 provided is the method of any preceding embodiment wherein the classification model uses expression levels for one or more species of one or more genus shown in Figure 9a.In embodiment 18 provided is the method of any preceding embodiment wherein the classification model uses expression levels for species of at least, exactly, and/or no more than any of 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30 of the genuses shown in Figure 9a. In embodiment 19 provided is the method of any preceding embodiment wherein a higher risk of developing or progressing glycemic dysregulation is determined, at least in part, from relatively lower expression of Ruminococus species compared to nondiabetic individuals. In embodiment 20 provided is the method of any preceding embodiment wherein the Ruminococus species comprise Ruminococcus bicirculans,

R. callidus, and/or R. champallensis . In embodiment 21 provided is the method of any preceding embodiment wherein a higher risk of developing or progressing glycemic dysregulation is determined, at least in part, from relatively lower expression of Blautia species in the sample compared to nondiabetic controls. In embodiment 22 provided is the method of any preceding embodiment wherein the Blautia species comprise Blautia massiliensis and/or Blautia Mar seille- P3087. In embodiment 23 provided is the method of any preceding embodiment wherein the classification model uses at least, exactly, and/or no more than any of 1, 2, 3, 4, 5, 6, 7, 8, 9, 10,

11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36,

37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62,

63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88,

89, 90, 91, 92, 93, 94, or 95 of the KOs selected from those of Figure 9b. In embodiment 24 provided is the method of any preceding embodiment further comprising outputting the inference to a user interface device or to a computer-readable memory. In embodiment 25 provided is a method comprising delivering and/or administering to an individual one or more therapeutic interventions to decrease glycemic dysregulation or the development of glycemic dysregulation, wherein the individual has been found to be at risk for developing or progressing glycemic dysregulation by the method of any preceding embodiment. In embodiment 26 provided is the method of embodiment 25 wherein the therapeutic intervention comprises administering one or more pharmaceutical agents to the individual. In embodiment 27 provided is the method of embodiment 25 or 26 wherein the one or more pharmaceutical agents comprise metformin, a sodium glucose co-tranporter 2 inhibitor (SGLT2i), a glucagon-like peptide 1 receptor agonist (GLP-RA), or a combination thereof. In embodiment 28 provided is the method of embodiment 27 wherein the pharmaceutical agent comprises metformin. In embodiment 29 provided is the method of embodiment 28 wherein 500 mg of metformin is taken daily. In embodiment 30 provided is the method of any preceding embodiment wherein 1000 mg of metformin is taken daily. In embodiment 31 provided is the method of embodiment 28 wherein 1500 mg of metformin is taken daily. In embodiment 32 provided is the method of embodiment 28 wherein 2000 mg of metformin is taken daily. In embodiment 33 provided is the method of embodiment 28 wherein 2500 mg of metformin is taken daily. In embodiment 34 provided is the method of any one of embodiments 26 through 33 wherein the pharmaceutical agent comprises a SGLT2i selected from the group consisting of canagliflozin, dapagliflozin, and empagliflozin. In embodiment 35 provided is the method of any one of embodiments 26 through 34 wherein the pharmaceutical agent comprises a GLP-RA selected from the group consisting of Dulaglutide, liraglutide, lixisenatide, Exenatide, Exenatide extended release, Semaglutide, injection, and Semaglutide, oral. In embodiment 36 provided is the method of any one of embodiments 25 through 35 wherein the therapy comprises a dietary therapy. In embodiment 37 provided is the method of embodiment 36 wherein the dietary therapy comprises eating a whole foods plant- based diet and reducing or eliminating meat; reducing or eliminating cheese consumption, eating the largest meal of the day at midday, or a combination thereof. In embodiment 38 provided is the method of embodiment 36 or 37 wherein the dietary therapy comprises eating a diet of an average of less than 200, 150, 100, or 50 grams of carbohydrate daily, periodically fasting for at least 16 hours, or a combination thereof. In embodiment 39 provided is the method of any one of embodiments 25 through 38 wherein the therapy comprises exercise. In embodiment 40 provided is the method of any one of embodiments 25 through 39 wherein the therapy comprises administration of one or more supplements. In embodiment 41 provided is the method of embodiment 40 wherein the supplements comprise one or more pre-biotics, one or more probiotics, or a combination thereof. In embodiment 42 provided is the method of embodiment 41 wherein the one or more prebiotics and/or probiotics are chosen to modulate the gut microbiome of the individual toward a biome that favors decreased glucose dysregulation.

[0037] In embodiment 43 provided is a method for determining whether an individual is able to control blood glucose with metformin comprising a) providing a biological sample from the individual; b) sequencing nucleic acids from the sample to provide sequence information, wherein the nucleic acids comprise mRNA; c) determining, from the sequence information one or more features of the nucleic acid sequences; d) executing by computer a classification model that infers, from one or more features in the feature set, a measure of likelihood of control of blood glucose with metformin for the individual. In embodiment 44 provided is the method of embodiment 43 wherein the one or more features comprise (1) determinations of KEGG- Orthologs (KOs) for gene-level activity within the sample. In embodiment 45 provided is the method of embodiment 43 or 44 wherein the determinations of KOs comprise at least 5, 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 85, 90, 95, 100, 110, 120, 130, 150, 170, or 200 different KOs and/or not more than 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 85, 90, 95, 100, 110, 120, 130, 150, 170, 200 or 300 KOs. In embodiment 46 provided is the method of any of embodiments 43-45 wherein the one or more features comprise (2) measures of active microbial species within the sample, wherein the one or more measures are included in a feature set. In embodiment 47 provided is the method of any of embodiments 43-46 wherein the measures of active microbial species include measures of species from at least 1, 2, 5, 7, 10, 12, 15, 17, 20, 22, 25, 30, 40, or 50 genuses and/or not more than 2, 5, 7, 10, 12, 15, 17, 20, 22, 25, 30, 40, 50 or 70 genuses. In embodiment 48 provided is the method of any of embodiments 43- 47 wherein the biological sample comprises a stool sample. In embodiment 49 provided is the method of any of embodiments 43-48 wherein the individual is a human. In embodiment 50 provided is the method of any of embodiments 43-49 wherein the individual has an HbAlc of less than 5.7% and is conventionally classified as nondiabetic. In embodiment 51 provided is the method of any of embodiments 43-50 wherein the individual has an HbAlc of 5.7-6.4% and is conventionally classified as prediabetic. In embodiment 52 provided is the method of any of embodiments 43-51 wherein the individual has an HbAlc of 6. 5% or greater and is conventionally classified as diabetic. In embodiment 53 provided is the method of any of embodiments 43-52 wherein the sequence information comprises metatranscriptomic information. In embodiment 54 provided is the method of any of embodiments 43-53 wherein the feature set used by the classification algorithm includes at least measures of activity of one or more microbial taxa. In embodiment 55 provided is the method of any of embodiments 43-54 wherein the feature set used by the classification algorithm includes at least measures of taxa involved in one or more metabolic pathways. In embodiment 56 provided is the method of any of embodiments 43-55 wherein the classification model uses expression levels for one or more species of one or more genuses shown in Figure 10c. In embodiment 57 provided is the method of any of embodiments 43-56 wherein the classification model uses expression levels for species of at least, exactly, and/or no more than any of 1, 2, 3, 4, or 5, of the genuses shown in Figure 10c. In embodiment 58 provided is the method of any of embodiments 43-57 wherein the classification model uses at least, exactly, and/or no more than any of 1, 2, 3, 4, 5, 6, 7, 8, 9, 10,

63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, or 81, of the KOs selected from those of Figure lOd. In embodiment 59 provided is the method of any of embodiments 43- 58 wherein 500 mg of metformin is taken daily. In embodiment 60the method of any of embodiments 43-59 wherein 1000 mg of metformin is taken daily. In embodiment 61 provided is the method of any of embodiments 43-60 wherein 1500 mg of metformin is taken daily. In embodiment 62 provided is the method of any of embodiments 43-61 wherein 2000 mg of metformin is taken daily. In embodiment 63 provided is the method of any of embodiments 43- 62 wherein 2500 mg of metformin is taken daily.

[0038] In embodiment 64 provided is a method comprising: a) providing biological samples from each of a first set of subjects, a second set of subjects, and a third set of subjects, wherein the biological samples comprise a stool samples, and wherein the first set of subjects have been diagnosed with type 2 diabetes, the second set of subjects have been diagnosed as prediabetic, and the third set of subjects have been diagnosed as nondiabetic; b) sequencing nucleic acids in the biological samples to provide sequence information; and c) performing a statistical analysis on the sequence information to produce a model that predicts a risk of developing or progressing glucose dysregulation in an individual. In embodiment 65 provided is the method of embodiment 64 wherein the statistical analysis comprises an analysis selected from correlational, Pearson correlation, Spearman correlation, chi-square, comparison of means (e.g., paired t-test, independent t-test, ANOVA) regression analysis (e.g., simple regression, multiple regression, linear regression, non-linear regression, logistic regression, polynomial regression stepwise regression, ridge regression, lasso regression, elasticnet regression) and non-parametric analysis (e.g., Wilcoxon rank-sum test, Wilcoxon sign-rank test, sign test). In embodiment 66 provided is a method comprising administering a therapy to normalize or improve blood glucose regulation to an individual found to be at risk of developing or progressing blood glucose dysregulation by the method of embodiment 64 or 65. In embodiment 67 provided is the method of embodiment 66 wherein the therapy comprises any of the therapies described for embodiment 1.

[0039] In embodiment 68 provided is a system comprising: (i) a computer comprising: (a) a processor; and (b) a memory storage module, coupled to the processor, the memory storing a module comprising: (1) nucleic acid sequence information from a biological sample from a subject comprising a gut microbiome;(2) a classification model which, based on values including the measurements, classifies the subject as at risk for developing or progressing glucose dysregulation, wherein the classification model is selected to have a sensitivity of at least 75%, at least 85% or at least 95%; and(3) computer executable instructions for implementing the classification model on the test data.

[0040] In embodiment 69 provided is a method for developing a computer model for inferring, from feature data, a risk of developing or progressing glucose dysregulation in a subject comprising: a) training a machine learning algorithm on a training data set, wherein the training data set comprises, for each of a plurality of subjects, (1) a class label classifying a subject as nondiabetic, prediabetic, or diabetic; and (2) feature data comprising quantitative measures for each of a plurality of features selected from gut microbiome transcriptome expression, and wherein the machine learning algorithm develops a model that infers a class label for a subject based on the feature data.

[0041] In embodiment 70 provided is a method that infers a risk of developing or progressing glucose dysregulation in a subject, the method comprising: (a) providing a data set comprising, for the subject, feature data for each of a plurality of features selected from gut microbiome transcriptome gene expression data and taxa activity data; and(b) executing a computer model on the data set to infer the risk of developing or progressing glucose dysregulation in the subject. [0042] In embodiment 71 provided is a software product comprising a computer readable medium in tangible form comprising machine executable code, which, when executed by a computer processor, infers a risk of developing or progressing glucose dysregulation in a subject by:(a) accessing a data set comprising, for a subject, feature data for each of a plurality of features selected from gut microbiome transcriptome gene expression data and taxa activity data; and(b) executing a computer model on the data set to infer the risk of developing or progressing glucose dysregulation in the subject.

[0043] In embodiment 72 provided is a method of providing therapy to decrease a risk of developing or progressing glucose dysregulation or to normalize glucose regulation in a subject comprising: (a) inferring the presence of a risk of developing or progressing glucose dysregulation in a subject according to a method as described herein; and(b) administering a therapeutic intervention to the subject effective to decrease the risk of developing or progressing glucose dysregulation or to normalize glucose regulation in the subject.

[0044] In embodiment 73 provided is a method for diagnosing and treating developing or progressing glucose dysregulation in a subject, the method comprising: (a) receiving from a subject a sample comprising a gut microbiome; (b) determining nucleic acid sequences of a microorganism component of the sample; (c) determining alignments of the nucleic acid sequence to reference nucleic acid sequences associated with developing or progressing glucose dysregulation; (d) generating a microbiome feature dataset for the subject based upon the alignments; (e) generating an inference of the risk of developing or progressing glucose dysregulation in the subject upon processing the microbiome feature dataset with an inference model derived from a population of subjects; and (1) at an output device associated with the subject, providing a therapy to the subject with the risk of developing or progressing glucose dysregulation upon processing the inference with a therapy model designed to treat the glucose dysregulation.

[0045] In embodiment 74 provided is a method comprising: (a) measuring, in a sample from a subject comprising an gut microbiome, activity of one or more biomarkers selected from Figure 9;(b) inferring, from the measurements, a risk of developing or progressing glucose dysregulation in the subject; and(c) delivering to the subject a therapeutic intervention effective to decrease the risk of developing or progressing glucose dysregulation or to normalize glucose regulation in the subject.

[0046] In embodiment 75. a method comprising: a) providing biological samples from each of a first set of subjects and a second set of subjects having glucose dysregulation and having been subject to a therapeutic intervention, wherein the biological samples comprise a gut microbiome and wherein the first set of subjects responded positively to the therapeutic intervention and the second set of subjects did not respond positively to the therapeutic intervention; b) sequencing nucleic acids in the biological samples to provide sequence information; and c) performing a statistical analysis on the sequence information to produce a model that infers having a positive response or lack of positive response to the therapeutic intervention in a subject.

[0047] In embodiment 76 provided is a method of treating a subject with glucose dysregulation comprising: (a) inferring that the subject will respond positively to each of one or more therapeutic interventions by executing a model on nucleic acid information from a biological sample from the subject comprising a gut microbiome; and(b) administering to the subject one or more therapeutic interventions to treat the glucose dysregulation. In embodiment 77 provided is the method of embodiment 76 wherein the therapeutic intervention comprises administration of metformin.

[0048] In embodiment 78 provided is a method comprising at a computer system comprising at least one processor and a memory, storing at least one program for execution by the at least one processor: a) obtaining nucleic acid sequence data in electronic form for a first plurality of nucleic acids from a biological sample of the subject, wherein the nucleic acid sequence data comprises sequences for 500, 1000, 2000, 5000, 7000, or 10,000 or more nucleic acids, preferably 1000 or more, more preferably 5000 or more, even more preferably 10,000 or more nucleic acids, in the plurality of nucleic acids; b) obtaining phenotypic data about the subject in electronic form, the phenotypic data comprising a plurality of responses; c) identifying, from the nucleic acid sequence data, a corresponding presence or absence of each respective functional activity condition in a plurality of functional activity conditions in the subject by: 1) determining a corresponding measure of transcriptional activity for each respective KEGG ortholog designation in a plurality of KEGG ortholog designations by matching each respective nucleic acid in the first plurality of nucleic acids to individual sequences of the plurality of KEGG ortholog designations, 2) determining a corresponding measure of transcriptional activity for each respective microbial taxa in a plurality of microbial taxa by matching each respective nucleic acid in the first plurality of nucleic acids to individual sequences of the plurality of taxa, 3) determining, for each respective functional activity category in the plurality of functional activity categories, a corresponding functional activity score based on a corresponding measure of transcriptional activity for a respective KEGG ortholog designation in the plurality of KEGG ortholog designations or a corresponding measure of transcriptional activity for a respective microbial taxa in the plurality of microbial taxa, and4) identifying, for each respective functional activity condition in the plurality of functional activity conditions, the corresponding presence or absence of the respective functional activity condition based on a corresponding functional activity score; d) identifying, from the phenotypic data from the subject, a corresponding presence or absence of, or degree of, each respective phenotypic condition in a plurality of phenotypic conditions in the subject by: 1) assigning a corresponding numerical value to each respective response in the plurality of responses comprising the phenotypic data, 2) determining, for each respective phenotypic condition in the plurality of phenotypic conditions, a corresponding phenotype score based on a corresponding numerical value for one or more respective response in the plurality of responses, and 3) identifying, for each respective phenotypic condition in the plurality of phenotypic conditions, the corresponding presence or absence of, or degree of, each respective phenotypic condition in the plurality of phenotypic conditions based on a corresponding phenotypic score; e) accessing a knowledge base that includes for each of a plurality of food items (i) a corresponding first desirability ranking, in a plurality of desirability rankings, of the food for each respective phenotypic condition in the plurality of phenotypic conditions present in the subject and (ii) a corresponding second desirability ranking, in the plurality of desirability rankings, of the food for each respective functional activity condition in the plurality of functional activity conditions present in the subject; 1) using a recommendation engine, executing logic to produce a corresponding recommendation for each respective food item in the plurality of food items for the subject based on the corresponding first desirability rankings and the corresponding second desirability rankings for the respective food item, thereby generating a plurality of food recommendations; and g) outputting the plurality of food recommendations to an electronic device accessible by the subject.

EXAMPLES Example 1

[0049] In this study, the gut microbiome of 53,970 individuals was utilized, using a gut metatranscriptome approach combined with multivariate and machine learning analyses, in an effort to answer several questions: [0050] 1. How does the microbiome change as an individual progresses from non-diabetes

(normal glucose homeostasis), to prediabetes, to T2D? And, more importantly, what are the microbiome functional changes that take place throughout this spectrum of disease? In this Example, we delineate the microbiome differences between individuals with non-diabetes and prediabetes and we compare these changes with the differences between the microbiomes of individuals with prediabetes and T2D. By establishing taxonomic and functional progression, we should be able to inform further target discovery for modulating the pathways of microbiome with medications, prebiotic, probiotic, or other supplement recommendations.

[0051] 2. Can the gut microbiome provide insight into the early diagnosis of either prediabetes or T2D? With advanced predictive modeling, we aimed to develop a metabolic disease risk score, based on the gut metatranscriptome, that can predict the probability of having insulin dysregulation before detecting high glycated hemoglobin (HbAlc), which is used as the standard-of-care diagnostic for prediabetes and diabetes conditions. By using this risk score for T2D, patients can then take action to either prevent or postpone further progression of disease. Our aim is to pick up early signals for the beginning of metabolic disease, not just the beginning of diagnosis of the same. A related question is to understand if people who follow food and/or other lifestyle recommendations would show improved microbiome diabetes risk scores?

[0052] 3. What are the microbiome correlates of response to current treatments for individuals with T2D? There are multiple medications that are used in the treatment of T2D. For individuals who have already been diagnosed with T2D (or those with a prediabetes condition receiving treatment in the form of medication), can we identify microbiome-related determinants of response to treatment? Most frequently, patients receive metformin, and we aim to determine the microbiome signature of controlling HbAlc while they are taking metformin. For the patients who are not able to control their HBAlc while on metformin, this signature may point to additional interventions that can assist with glucose control.

[0053] 4. For individuals who choose not to take treatment in the form of medications or insulin therapy (and instead rely on diet and lifestyle changes), can we determine if there are microbiome correlates of response to these non-medication treatments? Can these correlates inform how to modulate the microbiome with a positive treatment effect? In other words, are there microbiome pathways that can be modulated so as to achieve better glycemic control? If so, then regular surveillance of these changes should be aided by the microbiome readout over time. [0054] Study participant recruitment. The data analyzed for the purpose of this Example was obtained from Viome customers, who either completed a research Informed Consent Form (approved by a federally -accredited Institutional Review Board), or agreed to have their data analyzed in the terms and conditions, during the purchase of a gut microbiome test. All study data are de-identified; the laboratory, bioinformatics, and data science team members never had access to any personally identifiable information.

[0055] Sample collection and laboratory analysis. The metatranscriptomic method that we use is designed for large-scale population analysis of stool samples as described previously (Hatch et al., 2019), and included sample collection, ambient temperature sample preservation, total RNA extraction, physical removal of ribosomal RNAs (rRNAs), preparation of directional Illumina libraries, and Illumina sequencing. The stability of the RNA stabilizer was tested for up to 28 days at ambient temperature, including shipping.

[0056] Phenotype data collection The clinical phenotypes and medication status were labeled based on the answers to the questionnaire. The non-diabetic group includes presumed healthy individuals who are not reporting medical conditions as well as individuals who have reported comorbidities, while excluding anyone taking antibiotics within one month, proton pump inhibitor and/or acid suppressants, had abdominal surgery or with diseases that may affect gut microbiome (e.g. IBS, IBD, colon cancer). For this group we have matched the age, sex and BMI to those of prediabetic and T2D group. The first stratification is based on the type of diagnosis that patients report, including pre-diabetes and T2D. Within the prediabetic and T2D groups there are patients who receive treatment in the form of medications, as well as patients who are not on medications. We consider within the T2D group that subjects who have reported HbAlc <6.5 to be controlling diabetes with treatment, and the patients with >=6.5 to be not controlling diabetes with treatment.

[0057] Bioinformatics processing. Paired-end reads were mapped to a catalog of 53,660 microbial genome assemblies spanning archaea, bacteria, fungi, protozoa, and viruses (the complete genomes available in NCBI Reference Sequence Database were downloaded, and the GenBank sequence database was used for viral genomes.) Strain-level relative activities were computed from mapped reads via the expectation-maximization (EM) algorithm (Dempster et al., 1977). Relative activities at other levels of the taxonomic tree were then computed by aggregation according to the taxonomic rank. Relative activities for biological functions were computed by mapping paired-end reads to a catalog of 52,324,420 microbial genes, quantifying gene-level relative activities with the EM algorithm, and then aggregating gene-level activity by KEGG Ortholog (KO) annotation (Kanehisa & Goto, 2000). The identified and quantified active microbial species and KOs for each sample were then provided to the T2D classifier.

[0058] Descriptive statistical analysis. Standard statistical analyses described below were initially performed to analyze the differential expression of active microorganisms and active functions between. The data was transformed using the centered log ratio transformation (CLR) (Aitchison, 1982) after imputation of zero values using multiplicative replacement (Martin- Femandez et al., 2003). The age, sex and BMI of non-diabetic controls were matched to those of samples with T2D or pre-diabetics. Mann-Whitney U (MWU) test (FDR < 0.05) was used to evaluate differential expression and Chi-2 test (FDR < 0.05) to evaluate differential prevalence of species or KOs between different groups. It is important to note that this is a descriptive statistical test to analyze features independently for differential expression without taking into account the interactions among features, and is thus not suitable for the machine learning classification method (below).

[0059] Mapping KOs to functional categories for presentation. For KO visualization scatterplots, the Python module “Bio.KEGG” was used to take as input the KO name and return KO hierarchy at three different levels (level-2 corresponds to level B, level-3 corresponds to level C, and level-4 corresponds to level D in KEGG). Level 2 and level 3 annotation were assigned using EM algorithms (Dempster et al., 1977) with KOs weighted by absolute values coefficients or reversed p values in log scale.

[0060] Richness and diversity. Richness was estimated as the number of present species or KOs in each sample. Shannon-diversity was computed to evaluate the alpha diversity of samples. All comparisons between richness or Shannon-diversity were done with independent t-test.

[0061] Machine learning. The microbiome data was transformed using the centered log ratio transformation (CLR) (Aitchison, 1982) after imputation of zero values using multiplicative replacement (Marti n-Femandez et al., 2003). Machine-learned models using stool microbiome are Support Vector Machines. Other models were also tried, such as logistic regression and random forest, and the model giving the highest AUC was adopted. Hyperparameter optimization was done using a nested 5-fold cross-validation on the discovery cohort. The inner layer was used for feature selection, training and tuning. The top 20% features with the highest KW test effect size were selected. The outer layer was used for estimation of model performance. The hyperparameters were scored based on balanced accuracy and the final models were evaluated based on balanced accuracy and AUC. In the diagnostic classifiers, the non-diabetic controls were subsampled with matched age, sex and BMI to samples with T2D or pre-diabetics. The number of subsampled non-diabetic controls was three times that of samples with conditions. For pre-diabetic vs. T2D classifier and all the treatment classifiers, the two classes were not matched because of the small sample size.

[0062] The machine learning models take in multiple features at the same time. In the SVM model’s view, the contribution of each feature is more complex than that of descriptive analysis which takes one feature at a time. In the results, some models will pick more features enriched in non-diabetic samples. It does not necessarily mean that the model’s only predicting non-diabetic samples. Instead, these features are enriched in non-diabetic samples and were also features with lower expression in diabetic samples.

[0063] Table 1 shows the characteristics of our discovery (n=50,942) and validation (n=3,028) cohorts.

[0064] Phenotype stratification. Figure describes the clinical phenotypes based on the self-reported labels in this Example. The non-diabetic group includes presumed healthy individuals who are not reporting medical or mental conditions and are not actively taking medications, as well as individuals who have reported comorbidities or are taking medications. The non-diabetic group excludes anyone taking antibiotics within one month of sample collection, proton pump inhibitors and/or acid suppressants, had abdominal surgery or report diseases that may affect gut microbiome significantly (e.g. IBS, IBD, colon cancer). For this group age, sex and body mass index (BMI) were matched to those in the prediabetic and T2D groups.

[0065] The first stratification is based on the type of diagnosis that patients report, i.e., prediabetes and T2D. Within these two groups, there are individuals who received treatment in the form of medications, as well as individuals who are not on medications, and instead have diet and lifestyle changes. It is considered that within the T2D group that subjects who have reported HbAlc <6.5 to be able to control their HbAlc using medication, and the subjects with >=6.5 to not be able to control their HbAlc. The stratification in Figure 1 was used to analyze disease progression, diagnosis and medication control, using multivariate statistical analysis and machine learning.

[0066] Summary of analyses. Table 2 presents a summary of the comparison cohorts and relevant analyses related to disease progression and T2D risk score. The first row represents the main cohort comparison used as the basis for the T2D risk score, to be described in more detail below. The next three rows represent the analyses for disease progression. The last row presents the analysis for the obesity /BMI related microbial features.

Table 2. Summary of methods and results for progression of disease and diagnostics for type 2 diabetes (T2D)

*med, medicated; non-med, non medicated

^"T

Comparison cohorts ; Cohort Diff. Prev. Diff. Expr. ; Classification j Mean ; Mean Classifi ; matchin si feats si feats \ method ROC- balanced er e

[0067] Obesity-related microbial composition and functional relevance. One of the most frequently reported confounding factors for T2D disease progression is obesity. In this Example the contribution of obesity phenotype to microbiome was first assessed by applying a statistical Kruskall-Wallis (KW) model from 27,982 subjects in our cohort who had low BMI <25

(n=17159) and belong to a normal range, intermediate BMI>=25 and BMI <30 (n=8013) who belong to the overweight range and high BMI >=30 (n=2810) who belong to the obese range. It was observed that 5485 KOs and 680 species were statistically significant in a three-way KW analysis with FDR <0.05. In all further analyses, the observed obesity and non-obesity -related features were highlighted so that the unique microbiome associations of the type 2 diabetes state could be delineated.

[0068] Disease Progression Descriptive statistics. In total, 4380 taxa and 6797 functional KOs were identified in our cohort. Figure 2 presents the differential expression, for the most statistically significant features (using Kruskal -Wallis method after Benj amini-Hochberg FDR correction, p-value<=0.05). Each figure presents non-T2D patients (in blue) who have not received any medication vs. T2D or prediabetes patients (in red). Figure 3 presents the differential prevalence corresponding to the same comparison groups.

[0069] Shown in scatter plots : 75 out of 4380 unique differentially expressed microorganisms were observed when pre-diabetic is compared to non-diabetic (see Figure 3b), and 111 unique microorganisms were observed when T2D was compared to non-diabetic samples (see Figure 2a). 546 differentially expressed KOs were observed in pre-diabetic vs. non-diabetic (see Figure 3d) and 206 differentially expressed KOs in T2D vs non-diabetic subjects (see Figure 2b). In contrast with previous studies which are performed using 16s technology, in this Example a wider spectrum of differential expression was observed and functional modules inferred directly from the experimental observations. Shown in side-by-side scatter plots (see Figure 3): 97 statistically significant species out of 4380 unique active microorganisms were observed in pre-diabetic vs. non-diabetic (see Figure 3a left panel), and 78 unique microbial species were observed in T2D samples vs. non-diabetic (see Figure 3a right panel) using Chi-squared tests after Benj amini-Hochberg FDR correction, p-value<=0.05). In addition, the results from the SVM model are presented in Figure 4.

[0070] 677 prevalent KOs were observed in pre-diabetic vs. non-diabetic (see Figure 3c left panel) and 115 prevalent KOs in T2D vs non-diabetic samples (see Figure 3c right panel) [0071] Ruminococcus and Blautia genera. Results from differential prevalence and expression of bacteria belonging to Blautia and Ruminococcus, genera are singled out, which are reported in the literature as positively associated with type 2 diabetes. However, in the present metatranscriptomic data it was observed that Ruminococcus bicirculans, R. callidus, and R. champallensis have actually lower expression in T2D patients, and only sp. DSM 100440 has higher prevalence and expression in T2D patients (see Table 2A and Figure 5a). In the literature, there is a prevalent notion that at the genus level, Blautia is positively associated with T2D. In the present data, with much higher resolution, it is possible to observe that out of the 8 species, Blautia massiliensis and Blautia Marseille-P 3087 have lower prevalence and expression in T2D as opposed to the non-T2D while there are six other species that have indeed higher prevalence and expression in Type 2 Diabetes patients (see Table 3 and Figure 5b).

Table 2A

Table 3. Prevalence and expression of the Blautia species along stages of disease progression

[0072] Diagnostic Model and Risk Score. To pick up early signals of metabolic disease, statistical analysis was performed to derive features that are associated with all the subjects with T2D vs. the general population who may have other comorbidities and may be subjected to different types of treatments, and separately performed ML analysis to derive a risk score for T2D. As shown in Figure 6b, taxa and KO diversity decrease in the T2D patients (KO diversity is significant p=3.27e-07). The taxonomic composition using the differential prevalence (based on Chi-square method) is presented (see Figure 7c) and KOs (see Figure 7e) of the microbiomes of Type 2 Diabetes patients vs. individuals in the general population. In Figure 7d the differential expression of taxa is presented and in Figure 7f functional changes using differential expression based on Kruskal-Wallis method. It was observed that there is a great overlap between the highest prevalent taxa with the statistical test that was presented in Figure 2, as 65 species are in common (out of 266 species) with the statistically significant species from the general population vs T2D, and yet, there are 201 unique species, which is due to the fact that there is a very big variability across this general population where there are influences from different comorbidities as well as various types of medications taken. Similarly, it was observed that 99 (out of 338 species) that are differentially expressed are statistically significant both in the non-medicated vs. T2D as well as in the general population vs. T2D cohorts (taxa and KO overlaps are presented in Figure 7g-j). A general classifier was developed with 922 KOs and species which is informative to provide a risk score that can predict whether an individual has a high risk of being diagnosed with T2D using the gut microbiome (see Figure 8a and Figure 9a-b). As shown in Figure 8b, in the validation cohort (n=3028) the model can distinguish between type 2 diabetes (mean=62.81) and non-diabetic (mean=44.10) individuals (t-test, p=2.41e-47), as well as between prediabetic(mean=53.25) and non-diabetic (p=2.88e-24) and also between T2D vs. pre-diabetic individuals (p=6.76e-07). A weak correlation between the T2D score and BMI of the validation cohort was observed (Rho=0.16, p=5.29e-19) (see Figure 9c), which suggests that the BMI alone is not sufficient and a predictive model is still needed to predict the risk of developing T2D.

[0073] Metformin control of Type 2 Diabetes In Table 4 all the treatment related comparison cohorts and relevant analyses are presented. Due to the limited number of samples the statistical analysis does not yield results except in the metformin related analysis there are two statistically significant KOs, namely K01273 and K02760, which show higher differential prevalence in those unable to achieve control (p<0.05). However, strong signals from an SVM approach were obtained, yielding ROC-AUC 0.74 when patients who were taking metformin and were able to control their HbAlc (<6.5), were compared to patients who were unable to control their HbAlc (>=6.5) (see Figure 10a). Table 4

[0074] Figure 11 presents results in richness (see Figure 11a) and diversity (see Figure lib) between Metformin-treated-HBAlc controlled and the non-controlled groups, and higher and statistically significant differences in KO richness and species richness in the non-controlled group were observed. The scatter plots are presented in Figure 10.

[0075] Strong signals were also obtained when individuals who are not taking any medications and instead rely on diet and lifestyle changes and are able to control their HbAlc (<6.5), were compared to patients who are unable to control their HbAlc (>=6.5) (see Figure 10a). Figure 11 presents results in richness (see Figure 11c) and diversity (see Figure lid) between non-medicated-HbAlc controlled groups, and it was observed higher statistically significant level for species diversity (p=3.06e-02), for patients who are able to control HbAlc. The SVM approach yielded a classifier showing ROC-AUC 0.69 (see Figure 10a). Scater plots of the features are presented in Figure 10.

[0076] Taxonomical Shifts in the T2D Metatranscriptome The association between type 2 diabetes and an imbalanced gut microbial composition has been reported in the past many years, and studies have suggested that gut dysbiosis is a factor in the development of insulin resistance (Allin et al., 2015; Sircana et al., 2018). Dysbiosis in T2D has been characterized by a reduction of Firmicutes species and an increase in the typically low-abundance organisms such as Proteobacteria (Larsen et al., 2010). The statistical model used in this Example reveals several enriched Firmicutes features on both the prediabetes and T2D sides when compared to the presumed healthy population (Figure 2a, Figure 3a and 3b). Firmicutes enriched in the T2D samples include Erysipelatoclostridium ramosum, Blautia producta, and uncharacterized species from Anaerococcus, Blautia, Clostridia, Clostridiales, Coprobacillus, and Streptococcus. Although more research is required to elucidate their roles in T2D development, Blautia and Streptococcus have been positively linked to T2D-associated gut microbiota (Candela et al.,

2016; Egshatyan et al., 2016). On the other hand, although the number of differentially prevalent and expressed Proteobacteria features is comparable in the diseased and non-diabetic cohorts, the enrichment of specific genera in the diseased populations is evident, and such distinction becomes more pronounced when progressing from the preT2D to T2D microbiome. E. ramosum, an opportunistic and obesogenic pathogen that may promote intestinal glucose and fat absorption (Mandic et al., 2019; Woting et al., 2014), has also been shown to be T2D-enriched in a metagenomic study of a Chinese cohort (Qin et al., 2012). Higher prevalence and abundance of Eggerthella and Ralstonia features, including E. lenta and R. pickettii, are also indicated by the statistical model in the pre-T2D and T2D samples (Figure 2a, Figure 3a and 3b). Interestingly, E. lenta is another opportunistic pathogen reported in the Chinese metagenome-wide study to be T2D-enriched, while R. pickettii, a pathogen capable of causing bacteremia and hospital infection outbreaks, has been suggested as an aggravator of glucose intolerance in obesity (Udayappan et al., 2017). Taken together, these results underscore the prominence of opportunistic pathogens in a disrupted microbiome and their potentials as T2D progression markers.

[0077] Enrichment of SCFA and Succinate-Producing Taxa in Presumed Healthy Cohort Studies have demonstrated the multi-mechanistic role of short-chain fatty acids in contributing to intestinal homeostasis. In particular, butyrate and acetate have been studied for their beneficial effects on insulin secretion, energy metabolism, and glucose tolerance (Gao et al., 2009; De Vadder et al., 2014; Yamashita et al., 2007). The descriptive analysis herein identifies several butyrate producers such as Butyricimonas , Butyrivibrio, Coprococcus, and Odoribacter to be differentially expressed in the non-diabetic samples (Figure 2a & Figure 3b). The enrichment of acetate producers such as Alistipes, Ruminococcus , Odoribacter, and Paraprevotella in the non-diabetic cohort is also noted. The presence of these SCFA-producing species aligns well with previous findings, highlighting their significance in normal or improved metabolic health. In addition to the enrichment of SCFA-producing species, the descriptive model suggests the capacity to produce succinate, a precursor/intermediate for SCFA, as a distinguishing feature, as many succinate-producing bacteria are present at higher abundance in the non-diabetic population compared to either T2D or preT2D (Figure 2a, Figure 3a and 3b). The observed taxa classify under the phylum Bacteroidetes, which is known to harbor numerous succinate-producing species, and include Alislipes putredinis , Alistipes ihumii, Parabacteroides, Paraprevotella clara, Odoribacter splanchnicus , and Ruminococcus callidus (Rautio et al., 2003; Parker et al., 2020; Sakamoto & Benno, 2006; Morotomi et al., 2009; Hiippala et al., 2020; La Reau & Suen, 2018). While the linkage between succinate, dysbiosis, and inflammatory diseases such as IBD is recognized, succinate has also been suggested to have beneficial metabolic effects and play a role in glucose tolerance and insulin sensitivity (De Vadder et al., 2016; Femandez- Veledo & Vendrell, 2019). Microbiota-derived succinate has been shown to promote glucose homeostasis by serving as a substrate of intestinal gluconeogenesis (IGN), leading to the inhibition of hepatic glucose production (De Vadder et al., 2016). Additionally, esters of succinic acid have been shown to be metabolized into succinate intracellularly and exert insulinotropic effects (MacDonald & Fahien, 1988).

[0078] Bile-Resistant Bacteria Associated with Presumed Healthy Individuals Colonic bacteria are capable of modifying and deconjugating primary bile acids synthesized in hepatocytes into secondary bile acids, resulting in a pool of diversified bile forms in the intestines. The relationship between the chemical diversity of the bile acids, their signaling properties, and bile homeostasis in T2D patients has been demonstrated increasingly. From the descriptive analysis herein, the differential expression of bile-resistant microbes, such as Alistipes, Bilophila, Bacteroides, and Sutterella is observed in the presumed healthy cohort (Figure 2a & Figure 3a and 3b). The abundance of Alistipes, Bilophila and Bacteroides has been shown to increase in animal-based diets, which is presumably the result of a higher fat intake and therefore increased activity of bile secretion compared to plant-based diets (David et al., 2014). Two Bilophila features are consistently enriched on the expression and prevalence level in the non-diabetes population with respect to T2D patients (Figure 2a and 3a ). Notably, Bilophila wadsworthia utilizes taurine as an electron acceptor and its growth has been shown to be promoted by high-fat diets due to the increased availability of taurine-conjugated bile acids synthesized to facilitate lipid digestion (Devkota et al., 2012). Functionally, although the statistical model does not pinpoint any KOs in the pathways of secondary bile acid production, the machine learning model identified bile salt hydrolase (K01442) to be an important feature for the prediction of preT2D and T2D microbiome. Overall, the enrichment of bile-resistant bacteria may be the result of unrestricted diet in the presumed healthy cohort, altered bile acid homeostasis in T2D patients, or other feedback mechanisms underlying the T2D microbiome. [0079] Taxa Features in the Metformin Treatment Model Mounting evidence over the past several years suggests metformin as a major modifier of the gut microbiota, and metformin treatments have been shown to promote the growth of certain intestinal bacteria, notably Akkermansia muciniphila and Escherichia spp. (Elbere et al., 2018; Karlsson et al., 2013; Lee & Ko, 2014; Wu et al., 2017). The treatment classifiers provided herein present a different view by predicting microbes potentially important in discriminating between those who can and cannot control their HbAlc in response to metformin treatment (Figure 10). In general, the model predicts the importance of several Bacteroides and Parabacteroides species, and these genera have been shown to be among the most frequently altered groups in metformin treatments (Huang et al., 2020). Bacteroides thetaiotaomicron, an opportunistic pathogen and common inhabitant of a normal gut microbiome, is capable of fermenting a broad range of polysaccharides and considered by the model to be important in predicting those who cannot control their HbAlc levels. B. thetaiotaomicron has been shown to modulate inflammation by activating the Treg pathways, which play a role in reducing inflammatory response and improving insulin resistance (Hoffmann et al., 2016), but its ability to enhance the energy harvest and nutrient absorption of the host has also been demonstrated (Hooper et al., 2001; Samuel & Gordon, 2006). Similarly, B. eggerthii is identified as a potential predictor of non-responders. This microbe has been shown to increase in abundance in T2D microbiome and to aggravate colitis in mice models (Dziarski et al., 2016; Medina-Vera et al., 2019). Overall, we find that more research is needed to understand the taxa in both the responder and non-responder groups, as the features seem to present a complex picture of metformin efficacy.

[0080] Discussion Literature abounds with metagenomic studies over the past decade, and many genera have been associated with T2D (Gurung et al., 2020). This Example observes certain functions and species related to opportunistic pathogens, lipopolysaccharides (LPS), oxidative stress, and osmotic stress, as associated with T2D. Active signatures associated with the non-diabetic cohort, such as taxa capable of SCFA and succinate production, are also identified. The observations from the data are summarized in Figure 12 and in the following section.

[0081] Taxonomical Shifts in the T2D Metatranscriptome Studies have demonstrated the alterations of gut microbiota composition in T2D and suggested that gut dysbiosis is a factor in the development of insulin resistance (Allin et al., 2015; Sircana et al., 2018). The statistical model presented here reveals specific Firmicutes and Proteobacteria features enriched in both the prediabetes and T2D samples when compared to the presumed healthy population (Figure 2a, 3a and 3b). Although many of these taxa are uncharacterized, some have been linked to T2D in metagenomics studies while others are known opportunistic pathogens associated with altered metabolic phenotypes. E. ramosum, a feature observed in diseased cohorts herein, is an opportunistic bacterium that may promote intestinal absorption of glucose and fat in obesity (Mandic et al., 2019; Woting et al., 2014), has also been shown to be T2D-enriched in a metagenomic study of a Chinese cohort (Qin et al., 2012). Higher prevalence and abundance of Eggerthella and Ralstonia features, including E. lenta and R. pickettii, are also observed in the data herein. The opportunistic E. lenta has been reported to be enriched in the T2D microbiome, and R pickettii has been suggested to aggravate glucose intolerance in obesity (Qin et al., 2012; Udayappan et al., 2017). Overall, the taxonomical shifts in the diseased cohorts herein highlight the presence of opportunistic pathogens and their potentials as T2D progression markers. On the other hand, the differential expression of SCFA-producing taxa, succinate-producing taxa, and bile-resistant microorganisms is noted in the non-diabetic cohort, as described and discussed above.

[0082] Lipopolysaccharide-Associated and Proinflammatory Features One of the mechanisms through which the gut microbiota interacts with the host is through the production and shedding of LPS, which elicits a pro-inflammatory cascaded response of the immune system. Metabolic endotoxemia, as a result of LPS translocation across the intestinal barrier and into the bloodstream, causes inflammation and is intertwined with the development of insulin resistance (Cani et al., 2007). In this Example, this phenomenon is best highlighted by several LPS- producing species identified to be enriched in the T2D cohort, such as R. picketti and Escherichia albertii (Figure 2a & 3a). The fecal abundance of R. picketti has been shown to increase in obese patients with impaired glucose tolerance and type 2 diabetes, and the organism has been suggested to play a role in the development of glucose intolerance and increased inflammatory markers (Udayappan et al., 2017). E. albertii is an enteropathogen increasingly associated with diarrhea and gastroenteritis. In addition, the abundance of Escherichia spp. has been shown to increase as a result of metformin treatment (Forslund et al., 2015). Functionally, the model also identifies KOs involved in the biosynthesis of O-antigen, the exterior constituent of LPS and often contributing to virulence, and the biosynthesis, regulation, and export of LPS, particularly in the prediabetes population.

[0083] Overall Functional Characteristics Previous metagenome-based studies report the enrichment of cell motility and flagellar assembly pathways in non-diabetic individuals (Karlsson et al., 2013; Qin et al., 2012). This finding is corroborated by the analysis herein using metatranscriptomics, with bacterial motility proteins including those involved in chemotaxis and pilus assembly identified to be differentially expressed in the non-diabetic cohort, particularly in the comparison to the T2D cohort. The T2D samples herein are functionally enriched with activities from carbohydrate metabolism, glycerolipid metabolism, glycerophospholipid metabolism, transporters, and xenobiotics degradation. In addition, unique features such as quorum sensing and ribosome biogenesis are also differentially expressed in the T2D cohort (Figure 2b).

[0084] Features of Osmotic and Oxidative Stress in Prediabetes and T2D Functional analyses of T2D metagenomes using Chinese and European cohorts have revealed the enrichment of transporters for sugars, amino acids, and ions (Karlsson et al., 2013; Qin et al., 2012). In the sampled population herein, transporter activities, especially those of ions, are differentially prevalent and expressed in the microbiome of the diseased cohorts compared to that of the non-diabetes. In addition, the high prevalence of osmolyte transporters is observed within both the preT2D and T2D cohorts, although the signal particularly is strong in the former, where several KOs related to glycine betaine transport are identified. Elevated plasma tonicity is thought to be a risk factor for diabetes progression in patients with hyperglycemia (Stookey et al., 2004), while hyperosmotic stress has been shown in adipocytes to suppress insulin action through the serine phosphorylation of insulin receptor substrate-1 (IRS1), leading to cellular insulin resistance (Gual et al., 2003). It is therefore surmised that the prediabetic gut microbiome experiences local osmotic stress, and by actively uptaking and accumulating osmoprotective compounds such as glycine betaine in the cells, bacteria adapt to osmotic stress in response to osmolarity changes in the environment (Sleator & Hill, 2002). The active functions of the preT2D cohort largely mirror that of the T2D microbiome in terms of carbohydrate metabolism, lipid metabolism, and transporters. Additionally functions were identified associated with oxidative stress, including KOs involved in sequestering oxidative damage and glutathione- dependent redox chemistry, to have higher differential expression in preT2D. Oxidative stress can induce inflammation and is known to play a significant role in the progression of type 2 diabetes (Folli et al., 2011). Metabolic endotoxemia has also been shown to promote oxidative stress, while antibiotic treatment reduces its occurrence (P. D. Cani et al., 2008). Previously, metagenomic studies using Chinese and European cohorts have revealed the enrichment of oxidative stress-related functions in T2D patients (Karlsson et al., 2013; Qin et al., 2012); here, it is hypothesized that such stress response is discernible by the prediabetic stage at the metatranscriptomic level. [0085] Functional Features of the Metformin Treatment Model Changes in microbial functions are thought to play a role in mediating the beneficial effects of diabetes treatments, and metformin has been associated with increased activities in pathways such as SCFA production, lipopolysaccharide biosynthesis, sphingolipid and fatty acid metabolism, transporters, amino acid biosynthesis and metabolism, and pyruvate metabolism (Forslund et al., 2015; Lee & Ko, 2014; Ma et al., 2018; Wu et al., 2017). The functional findings from the metformin treatment model are summarized in Figure 10b and as follows. The analysis of taxa features is as described above.

[0086] Adding to the observation that amino acid metabolism is altered in the microbiome of metformin-treated patients, the model further identifies features in cysteine and methionine metabolism in predicting control over HbAlc. Many of such features are involved in the metabolic reactions surrounding cysteine, a precursor to glutathione. Glutathione is one of the most critical molecules in the defense against oxidative stress. Low plasma cysteine level and the resultant decreased glutathione synthesis have been associated with inflammation in IBD patients (Sido et al., 1998). Perhaps in line with the importance of cysteine, several KOs surrounding the serine and glycine nodes are also suggested by the model. These amino acids serve as the precursors not only to each other but also to cysteine. Additionally, glycine is tied to increased insulin sensitivity (Adeva-Andany et al., 2018), and earlier studies have shown that glycine degradation activities are enriched in metformin-untreated T2D patients and in prediabetic exercise non-responders (Forslund et al., 2015; Liu et al., 2020). The treatment model provided herein also identifies some features in lysine biosynthesis to be important. The butyrogenic property of lysine has been demonstrated by a human gut commensal (Bui et al., 2015), but more information is required to clarify how microbially produced lysine is linked to T2D.

[0087] An earlier metagenomic study has demonstrated the reduced pyruvate synthase capacity in metformin-untreated T2D samples (Forslund et al., 2015). In another study combining fecal metagenomic data from T2D patients and metatranscriptomic data from in vitro gut simulators, pyruvate metabolism has been found to be enriched with metformin treatment in both types of analysis (Wu et al., 2017). From the model presented herein, KOs surrounding the pyruvate node are identified to be important; along with several features involved in the TCA cycle, the importance of the central carbon metabolism is thus highlighted. Although it is unclear how these features contribute to metformin response, a possible explanation may be the accompanying production of SCFAs or organic acids such as succinate from pyruvate fermentation. The importance of energy metabolism and nucleotide biosynthesis in metformin- responders is also highlighted by multiple KOs from the TCA cycle, oxidative phosphorylation, and pentose phosphate pathway (Figure 10b). [0088] Applications include a stool test that can predict whether metformin will work for a particular individual. Furthermore, the gut microbiome can potentially be modulated to make metformin more effective.

Conclusions

[0089] The richness of the metatranscriptomic data also enabled development and validation of a T2D risk model, with a score that can distinguish individuals with glycemic dysregulation (prediabetes and diabetes) from those with normal glucose metabolism.

[0090] The methodologies include clinical grade laboratory analyses and bioinformatics, machine learning, and advanced statistical approaches leveraging data from 53,947 individuals. While some o findings are consistent with the published literature, surprisingly, novel associations were observed, as the metatranscriptome approach allows for more direct functional observations of the human gut microbiome.

[0091] The metatranscriptomic analysis illuminates the significance of both the taxa and microbial pathways pertaining to opportunistic pathogens, lipopolysaccharides (LPS), lipid metabolism, and ion transporters, as well as implications for inflammation, oxidative stress, and osmotic stress in the diseased cohorts. In the non-diabetic, presumed-healthy population, the analysis highlights the elevated activity of features associated with bile resistance, short chain fatty acids (SCFAs) production, cell motility, and succinate production that have not been reported before at the metatranscriptomic level.

[0092] Additionally, the machine learning model was able to distinguish novel metatranscriptomic features that segregate patients who receive metformin and are able to control their HbAlc from those who do not. The model points to complex changes in the amino acid metabolism, pyruvate metabolism, TCA cycle, and oxidative phosphorylation, as well as proteins related to genetic information processing important in determining treatment response.

[0093] References https://doi.org/10.1016/S0140-6736(16)00618-8

Adeva-Andany, M., Souto-Adeva, G., Ameneiros-Rodriguez, E., Femandez-Femandez, C, Donapetiy-Garcia, C., & Dominguez-Montero, A. (2018). Insulin resistance and glycine metabolism in humans. Amino Acids, 50(1), 11-27. https://doi.org/10.1007/s00726-017-2508-0

Aitchison, J. (1982). The Statistical Analysis of Compositional Data. Journal of the Royal Statistical Society: Series B (Methodological), 44(2), 139-160. https://doi.Org/10.llll/j.2517-6161.1982.tb01195.x Allin, K. H., Nielsen, T., & Pedersen, O. (2015). Mechanisms in endocrinology: Gut microbiota inpatients with type 2 diabetes mellitus. European Journal of Endocrinology, 172(4), R167-177. https://doi.org/10.1530/EJE- 14-0874

Bailes, B. K. (2002). Diabetes mellitus and its chronic complications. AORN Journal, 76(2), 266-276, 278-282; quiz 283-286. https://doi.org/10.1016/s0001-2092(06)61065-x Bui, T. P. N., Ritari, J., Boeren, S., de Waard, P., Plugge, C. M., & de Vos, W. M. (2015). Production of butyrate from lysine and the Amadori product fructoselysine by a human gut commensal. Nature Communications, 6(1), 10062. https://doi.org/10.1038/ncommsl0062

Cani, P. D., Bibiloni, R., Knauf, C., Waget, A., Neyrinck, A. M., Delzenne, N. M., & Burcelin, R. (2008). Changes in Gut Microbiota Control Metabolic Endotoxemia-Induced Inflammation in High-Fat Diet-Induced Obesity and Diabetes in Mice. Diabetes, 57(6), 1470-1481. https://doi.org/10.2337/db07-1403 Cani, Patrice D., Amar, J., Iglesias, M. A., Poggi, M., Knauf, C., Bastelica, D., Neyrinck, A. M., Fava, F., Tuohy, K. M., Chabo, C., Waget, A., Delmee, E., Cousin, B., Sulpice, T., Chamontin, B., Ferrieres, J., Tanti, J.-F., Gibson, G. R., Casteilla, L., ... Burcelin, R. (2007). Metabolic Endotoxemia Initiates Obesity and Insulin Resistance. Diabetes, 56(7), 1761-1772. https://doi.org/10.2337/db06-1491 Chatteqee, S., Khunti, K., & Davies, M. J. (2017). Type 2 diabetes. Lancet (London, England), 389(10085), 2239- 2251. https://doi.org/10.1016/S0140-6736(17)30058-2

Dempster, A. P., Laird, N. M., & Rubin, D. B. (1977). Maximum Likelihood from Incomplete Data Via the EM Algorithm. Journal of the Royal Statistical Society: Series B (Methodological) , 39(1), 1-22. https://doi.org/10.111 l/j.2517-6161.1977 tb01600.x

Folli, F., Corradi, D., Fanti, P., Davalli, A., Paez, A., Giaccari, A., Perego, C., & Muscogiuri, G. (2011). The role of oxidative stress in the pathogenesis of type 2 diabetes mellitus micro- and macrovascular complications: Avenues for a mechanistic -based therapeutic approach. Current Diabetes Reviews, 7(5), 313-324. https://doi.org/10.2174/157339911797415585

Forslund, K., Flildebrand, F., Nielsen, T., Falony, G., Le Chatelier, E., Sunagawa, S., Prifti, E., Vieira-Silva, S., Gudmundsdottir, V., Pedersen, H. K., Arumugam, M., Kristiansen, K., Voigt, A. Y., Vestergaard, H., Flercog, R., Costea, P. L, Kultima, J. R., Li, J., Jorgensen, T., ... Pedersen, O. (2015). Disentangling type 2 diabetes and metformin treatment signatures in the human gut microbiota. Forslund, K., Hildebrand, F., Nielsen, T., Falony, G., Le Chatelier, E., Sunagawa, S., Prifti, E., Vieira-Silva, S., Gudmundsdottir, V., Pedersen, H. K., Arumugam, M., Kristiansen, K., Voigt, A. Y., Vestergaard, H., Hercog, R., Costea, P. L, Kultima, J. R., Li, J., Jorgensen, T., ... Pedersen, O. (2015). Disentangling type 2 diabetes and metformin treatment signatures in the human gut microbiota. Nature, 528(7581), 262-266. https://doi.org/10.1038/naturel5766

Gual, P., Gonzalez, T., Gremeaux, T., Barres, R., Le Marchand-Brustel, Y., & Tanti, J.-F. (2003). Hyperosmotic stress inhibits insulin receptor substrate-1 function by distinct mechanisms in 3T3-L1 adipocytes. The Journal of Biological Chemistry, 278(29), 26550-26557. https://doi.org/10.1074/jbc.M212273200 Gurung, M., Li, Z., You, H., Rodrigues, R., Jump, D. B., Morgun, A., & Shulzhenko, N. (2020). Role of gut microbiota in type 2 diabetes pathophysiology. EBioMedicine, 51, 102590. https://doi.Org/10.1016/j.ebiom.2019.ll.051

Hatch, A., Home, J., Toma, R., Twibell, B. L., Somerville, K. M., Pelle, B., Canfield, K. P., Genkin, M., Banavar,

G., Perlina, A., Messier, H., Klitgord, N., & Vuyisich, M. (2019). A Robust Metatranscriptomic Technology for Population-Scale Studies of Diet, Gut Microbiome, and Human Health. International Journal of Genomics, 2019. https://doi.org/10.1155/2019/1718741 Hu, F. B. (2011). Globalization of Diabetes: The role of diet, lifestyle, and genes. Diabetes Care, 34(6), 1249-1257. https://doi.org/10.2337/dcll-0442

Kanehisa, M., & Goto, S. (2000). KEGG: Kyoto encyclopedia of genes and genomes. Nucleic Acids Research,

28(1), 27-30. https://doi.org/10.1093/nar/28.L27

Karlsson, F. H., Tremaroli, V., Nookaew, L, Bergstrom, G., Behre, C. J., Fagerberg, B., Nielsen, J., & Backhed, F. (2013). Gut metagenome in European women with normal, impaired and diabetic glucose control. Nature, 798(7452), 99-103. https://doi.org/10.1038/naturel2198

Larsen, N., Vogensen, F. K., van den Berg, F. W. J., Nielsen, D. S., Andreasen, A. S., Pedersen, B. K., Al-Soud, W. A., Sorensen, S. J., Hansen, L. H., & Jakobsen, M. (2010). Gut microbiota in human adults with type 2 diabetes differs from non-diabetic adults. PloS One, 5(2), e9085. https://doi.org/10.1371/joumal.pone.0009085

Lawlor, N., George, J., Bolisetty, M., Kursawe, R., Sun, L., Sivakamasundari, V., Kycia, L, Robson, P., & Stitzel,

M. L. (2017). Single-cell transcriptomes identify human islet cell signatures and reveal cell-type-specific expression changes in type 2 diabetes. Genome Research, 27(2), 208-222. https://doi.org/10.1101/gr.212720.116

Lee, H., & Ko, G. (2014). Effect of metformin on metabolic improvement and gut microbiota. Applied and Environmental Microbiology, 80(19), 5935-5943. https://doi.org/10.1128/AEM.01357-14 Ley, R. E., Turnbaugh, P. J., Klein, S., & Gordon, J. I. (2006). Microbial ecology: Human gut microbes associated with obesityLey, R. E., Turnbaugh, P. J., Klein, S., & Gordon, J. I. (2006). Microbial ecology: Human gut microbes associated with obesity. Nature, 444(1122), 1022-1023. https://doi.org/10.1038/4441022a Li, Q., Chang, Y., Zhang, K., Chen, H., Tao, S., & Zhang, Z. (2020). Implication of the gut microbiome composition of type 2 diabetic patients from northern China. Scientific Reports, 10(1), 5450. https://doi.org/10.1038/s41598-020-62224-3

Liu, Y., Wang, Y., Ni, Y., Cheung, C. K. Y., Lam, K. S. L., Wang, Y., Xia, Z., Ye, D., Guo, J., Tse, M. A.,

Panagiotou, G., & Xu, A. (2020). Gut Microbiome Fermentation Determines the Efficacy of Exercise for Diabetes Prevention. Cell Metabolism, 37(1), 77-91. e5. https://doi.Org/10.1016/j.cmet.2019.ll.001 Ma, W., Chen, J., Meng, Y., Yang, J., Cui, Q., & Zhou, Y. (2018). Metformin Alters Gut Microbiota of Healthy Mice: Implication for Its Potential Role in Gut Microbiota Homeostasis. Frontiers in Microbiology, 9,

1336. https://doi.org/10.3389/fmicb.2018.01336

Mandic, A. D., Woting, A., Jaenicke, T., Sander, A., Sabrowski, W., Rolle-Kampcyk, U., von Bergen, M., & Blaut, M. (2019). Clostridium ramosum regulates enterochromaffin cell development and serotonin release. Scientific Reports, 9(1), 1177. https://doi.org/10.1038/s41598-018-38018-z Martin-Femandez, J. A., Barcelo-Vidal, C, & Pawlowsky-Glahn, V. (2003). Dealing with Zeros and Missing

Values in Compositional Data Sets Using Nonparametric Imputation. Mathematical Geology, 35(3), 253- 278. https://doi.org/10.1023/A: 1023866030544

Qin, J., Li, Y., Cai, Z., Li, S., Zhu, J., Zhang, F., Liang, S., Zhang, W., Guan, Y., Shen, D., Peng, Y., Zhang, D., Jie, Z., Wu, W., Qin, Y., Xue, W., Li, J., Flan, L., Lu, D., ... Wang, J. (2012). A metagenome-wide association study of gut microbiota in type 2 diabetes. Nature, 490(7418), 55-60. https://doi.org/10.1038/naturell450 Ridaura, V. K., Faith, J. J., Rey, F. E., Cheng, J., Duncan, A. E., Kau, A. L., Griffin, N. W., Lombard, V., Flenrissat, B., Bain, J. R., Muehlbauer, M. J., Ilkayeva, O., Semenkovich, C. F., Funai, K., Flayashi, D. K., Lyle, B. J., Martini, M. C., Ursell, L. K., Clemente, J. C., ... Gordon, J. I. (2013). Gut microbiota from twins discordant for obesity modulate metabolism in mice. Science (New York, N.Y.), 347(6150), 1241214. https://doi.org/10.1126/science.1241214

Segerstolpe, A., Palasantza, A., Eliasson, P., Andersson, E.-M., Andreasson, A.-C., Sun, X., Picelli, S., Sabirsh, A., Clausen, M., Bjursell, M. K., Smith, D. M., Kasper, M., Ammala, C., & Sandberg, R. (2016). Single-Cell Transcriptome Profiling of Human Pancreatic Islets in Flealth and Type 2 Diabetes. Cell Metabolism, 24(4), 593-607. https://doi.Org/10.1016/j.cmet.2016.08.020

Sengupta, U., Ukil, S., Dimitrova, N., & Agrawal, S. (2009). Expression-based network biology identifies alteration in key regulatory pathways of type 2 diabetes and associated risk/complications. PloS One, 4(12), e8100. https://doi.org/10.1371/joumal.pone.0008100

Sido, B., Flack, V., Flochlehnert, A., Lipps, H., Flerfarth, C., & Droge, W. (1998). Impairment of intestinal glutathione synthesis inpatients with inflammatory bowel disease. Gut, 42(4), 485-492. https://doi.org/10.1136/gut.42.4.485

Sircana, A., Framarin, L., Leone, N., Berrutti, M., Castellino, F., Parente, R., De Michieli, F., Paschetta, E., & Musso, G. (2018). Altered Gut Microbiota in Type 2 Diabetes: Just a Coincidence? Current Diabetes Reports, 78(10), 98. https://doi.org/10.1007/sll892-018-1057-6 Sleator, R. D., & Hill, C. (2002). Bacterial osmoadaptation: The role of osmolytes in bacterial stress and virulence.

FEMS Microbiology Reviews, 26(1), 49-71. https://doi.Org/10.llll/j.1574-6976.2002.tb00598.x Stookey, J. D., Pieper, C. F., & Cohen, H. J. (2004). Hypertonic hyperglycemia progresses to diabetes faster than normotonic hyperglycemia. European Journal of Epidemiology, 79(10), 935-944. https://doi.org/10.1007/sl0654-004-5729-y

Turnbaugh, P. J., Hamady, M., Yatsunenko, T., Cantarel, B. L., Duncan, A., Ley, R. E., Sogin, M. L., Jones, W. J., Roe, B. A., Affourtit, J. P., Egholm, M., Henrissat, B., Heath, A. C., Knight, R., & Gordon, J. I. (2009). A core gut microbiome in obese and lean twins. Nature, 457(7228), 480-484. https://doi.org/10.1038/nature07540

Udayappan, S. D., Kovatcheva-Datchary, P., Bakker, G. J., Havik, S. R., Herrema, H., Cani, P. D., Bouter, K. E., Belzer, C., Witjes, J. J., Vrieze, A., de Sonnaville, E. S. V., Chaplin, A., vanRaalte, D. H., Aalvink, S., Dallinga-Thie, G. M., Heilig, H. G. H. J., Bergstrom, G., van der Meij, S., van Wagensveld, B. A., ... Nieuwdorp, M. (2017). Intestinal Ralstonia pickettii augments glucose intolerance in obesity. PloS One, 12(11), e0181693. https://doi.org/10.1371/joumal.pone.0181693 Wang, H., Lu, Y., Yan, Y., Tian, S., Zheng, D., Leng, D., Wang, C., Jiao, J., Wang, Z., & Bai, Y. (2020). Promising Treatment for Type 2 Diabetes: Fecal Microbiota Transplantation Reverses Insulin Resistance and Impaired Islets. Frontiers in Cellular and Infection Microbiology, 9. https://doi.org/10.3389/fcimb.2019.00455 Wellen, K. E., & Hotamisligil, G. S. (2005). Inflammation, stress, and diabetes. The Journal of Clinical Investigation, 775(5), 1111-1119. https://doi.org/10.1172/JCI25102 Woting, A., Pfeiffer, N., Loh, G., Klaus, S., & Blaut, M. (2014). Clostridium ramosum promotes high-fat diet- induced obesity in gnotobiotic mouse models. MBio, 5(5), e01530-01514. https://doi.org/10.1128/mBio.01530-14

Wu, H., Esteve, E., Tremaroli, V., Khan, M. T., Caesar, R., Manneras-Holm, L., Stahlman, M., Olsson, L. M.,

Serino, M., Planas-Felix, M., Xifra, G., Mercader, J. M., Torrents, D., Burcelin, R., Ricart, W., Perkins, R., Fernandez-Real, J. M., & Backhed, F. (2017). Metformin alters the gut microbiome of individuals with treatment -naive type 2 diabetes, contributing to the therapeutic effects of the drag. Wu, H., Esteve, E., Tremaroli, V., Khan, M. T., Caesar, R., Manneras-Holm, L., Stahlman, M., Olsson, L. M., Serino, M., Planas-Felix, M., Xifra, G., Mercader, J. M., Torrents, D., Burcelin, R., Ricart, W., Perkins, R., Fernandez- Real, J. M., & Backhed, F. (2017). Metformin alters the gut microbiome of individuals with treatment- naive type 2 diabetes, contributing to the therapeutic effects of the drag. Nature Medicine, 23(1), 850-858. https://doi.org/10.1038/nm 4345

Zhang, X., Shen, D., Fang, Z., Jie, Z., Qiu, X., Zhang, C., Chen, Y., & Ji, L. (2013). Human gut microbiota changes reveal the progression of glucose intolerance. PloS One, 8(8), e71108. https://doi.org/10.1371/joumal.pone.0071108

Zhou, B., Lu, Y., Hajifathalian, K, Bentham, J., Cesare, M. D., Danaei, G., Bixby, H., Cowan, M. J., Ali, M. K.,

Taddei, C., Lo, W. C., Reis-Santos, B., Stevens, G. A., Riley, L. M., Miranda, J. J., Bjerregaard, P., Rivera, J. A., Fouad, H. M., Ma, G., ... Cisneros, J. Z. (2016). Worldwide trends in diabetes since 1980: A pooled analysis of 751 population-based studies with 4-4 million participants. The Lancet, 387(10027), 1513—

1530. https://doi.org/10.1016/SO 140-6736(16)00618-8 Allin, K. H., Nielsen, T., & Pedersen, O. (2015). Mechanisms in endocrinology: Gut microbiota in patients with type 2 diabetes mellitus. European Journal of Endocrinology, 172(4), R167-177. https://doi.org/10.1530/EJE-14-0874 Candela, M., Biagi, E., Soverini, M., Consolandi, C., Quercia, S., Severgnini, M., Peano, C., Turroni, S., Rampelli, S., Pozzilli, P., Pianesi, M., Fallucca, F., & Brigidi, P. (2016). Modulation of gut microbiota dysbioses in type 2 diabetic patients by macrobiotic Ma-Pi 2 diet. The British Journal of Nutrition, 116(1), 80-93. https://doi.org/10.1017/S0007114516001045

David, L. A., Maurice, C. F., Carmody, R. N., Gootenberg, D. B., Button, J. E., Wolfe, B. E., Ling, A. V., Devlin, A. S., Varma, Y., Fischbach, M. A., Biddinger, S. B., Dutton, R. J., & Turnbaugh, P. J. (2014). Diet rapidly and reproducibly alters the human gut microbiome. Nature, 505(7484), 559-563. https://doi.org/10.1038/naturel2820

De Vadder, F., Kovatcheva-Datchary, P., Zitoun, C., Duchampt, A., Backhed, F., & Mithieux, G. (2016).

Microbiota-Produced Succinate Improves Glucose Flomeostasis via Intestinal Gluconeogenesis. Cell Metabolism, 24(1), 151-157. https://doi.Org/10.1016/j.cmet.2016.06.013 De Vadder, F., Kovatcheva-Datchary, P., Goncalves, D., Vinera, J., Zitoun, C., Duchampt, A., Backhed, F., &

Mithieux, G. (2014). Microbiota-Generated Metabolites Promote Metabolic Benefits via Gut-Brain Neural Circuits. Cell, 156(1), 84-96. https://doi.Org/10.1016/j.cell.2013.12.016 Devkota, S., Wang, Y., Musch, M. W., Leone, V., Fehlner-Peach, FI., Nadimpalli, A., Antonopoulos, D. A., Jabri,

B., & Chang, E. B. (2012). Dietary -fat-induced taurocholic acid promotes pathobiont expansion and colitis in 1110-/- mice. Nature, 487(1405), 104-108. https://doi.org/10.1038/naturell225 Dziarski, R., Park, S. Y., Kashyap, D. R., Dowd, S. E., & Gupta, D. (2016). Pglyrp-Regulated Gut Microflora

Prevotella falsenii, Parabacteroides distasonis and Bacteroides eggerthii Enhance and Alistipes finegoldii Attenuates Colitis in Mice. PloS One, 11(1), e0146162. https://doi.org/10.1371/joumal.pone.0146162 Egshatyan, L., Kashtanova, D., Popenko, A., Tkacheva, O., Tyakht, A., Alexeev, D., Karamnova, N., Kostryukova, E., Babenko, V., Vakhitova, M., & Boytsov, S. (2016). Gut microbiota and diet inpatients with different glucose tolerance. Endocrine Connections, 5(1), 1-9. https://doi.org/10. 1530/EC- 15-0094 Elbere, L, Kalnina, L, Silamikelis, L, Konrade, L, Zaharenko, L., Sekace, K., Radovica-Spalvina, L, Fridmanis, D., Gudra, D., Pirags, V., & Klovins, J. (2018). Association of metformin administration with gut microbiome dysbiosis in healthy volunteers. PLoS ONE, 13(9). https://doi.org/10.1371/joumal.pone.0204317 Femandez-Veledo, S., & Vendrell, J. (2019). Gut microbiota-derived succinate: Friend or foe in human metabolic diseases? Reviews in Endocrine & Metabolic Disorders, 20(4), 439-447. https://doi.org/10.1007/sl 1154- 019-09513 -z

Gao, Z., Yin, J., Zhang, J., Ward, R. E., Martin, R. J., Lefevre, M., Cefalu, W. T., & Ye, J. (2009). Butyrate

Improves Insulin Sensitivity and Increases Energy Expenditure in Mice. Diabetes, 58(1), 1509-1517. https://doi.org/10.2337/db08-1637

Fliippala, K., Barreto, G., Burrello, C., Diaz-Basabe, A., Suutarinen, M., Kainulainen, V., Bowers, J. R., Lemmer,

D., Engelthaler, D. M., Eklund, K. K., Facciotti, F., & Satokari, R. (2020). Novel Odoribacter splanchnicus Strain and Its Outer Membrane Vesicles Exert Immunoregulatoiy Effects in vitro. Frontiers in Microbiology, 11, 575455. https://doi.org/10.3389/fmicb.2020.575455 Floffmann, T. W., Pham, H.-P., Bridonneau, C., Aubiy, C., Lamas, B., Martin-Gallausiaux, C., Moroldo, M.,

Rainteau, D., Lapaque, N., Six, A., Richard, M. L., Fargier, E., Le Guem, M.-E., Langella, P., & Sokol, FI. (2016). Microorganisms linked to inflammatoiy bowel disease-associated dysbiosis differentially impact host physiology in gnotobiotic mice. The ISME Journal, 10(2), 460-477. https://doi.org/10.1038/ismej.2015.127

Flooper, L. V., Wong, M. FI., Thelin, A., Flansson, L., Falk, P. G., & Gordon, J. I. (2001). Molecular analysis of commensal host-microbial relationships in the intestine. Science (New York, N.Y.), 291(5505), 881-884. https://doi.org/10.1126/science.291.5505.881

Fluang, X., Flong, X., Wang, J., Sun, T., Yu, T., Yu, Y., Fang, J., & Xiong, FI. (2020). Metformin elicits antitumour effect by modulation of the gut microbiota and rescues Fusobacterium nucleatum-induced colorectal tumourigenesis. EBioMedicine, 61, 103037. https://doi.Org/10.1016/j.ebiom.2020.103037 Karlsson, F. FI., Tremaroli, V., Nookaew, F, Bergstrom, G., Behre, C. J., Fagerberg, B., Nielsen, J., & Backhed, F. (2013). Gut metagenome in European women with normal, impaired and diabetic glucose control. Nature, 498(1452), 99-103. https://doi.org/10.1038/naturel2198

LaReau, A. J., & Suen, G. (2018). The Ruminococci: Key symbionts of the gut ecosystem. Journal of Microbiology (Seoul, Korea), 56(3), 199-208. https://doi.org/10.1007/sl2275-018-8024-4 Larsen, N., Vogensen, F. K., van den Berg, F. W. J., Nielsen, D. S., Andreasen, A. S., Pedersen, B. K., Al-Soud, W. A., Sorensen, S. J., Flansen, L. FI., & Jakobsen, M. (2010). Gut microbiota in human adults with type 2 diabetes differs from non-diabetic adults. PloS One, 5(2), e9085. https://doi.org/10.1371/joumal.pone.0009085

Lee, FI., & Ko, G. (2014). Effect of metformin on metabolic improvement and gut microbiota. Applied and Environmental Microbiology, 80(19), 5935-5943. https://doi.org/10.1128/AEM.01357-14 MacDonald, M. J., & Fahien, L. A. (1988). Glyceraldehyde phosphate and methyl esters of succinic acid. Two “new” potent insulin secretagogues. Diabetes, 37(7), 997-999. https://doi.Org/10.2337/diab.37.7.997 Mandic, A. D., Woting, A., Jaenicke, T., Sander, A., Sabrowski, W., Rolle-Kampcyk, U., von Bergen, M., & Blaut, M. (2019). Clostridium ramosum regulates enterochromaffin cell development and serotonin release. Scientific Reports, 9(1), 1177. https://doi.org/10.1038/s41598-018-38018-z Medina-Vera, L, Sanchez-Tapia, M., Noriega-Lopez, L., Granados-Portillo, O., Guevara-Cmz, M., Flores-Lopez,

A., Avila-Nava, A., Fernandez, M. L., Tovar, A. R., & Torres, N. (2019). A dietary intervention with functional foods reduces metabolic endotoxaemia and attenuates biochemical abnormalities by modifying faecal microbiota in people with type 2 diabetes. Diabetes & Metabolism, 45(2), 122-131. https://doi.Org/10.1016/j.diabet.2018.09.004

Morotomi, M., Nagai, F., Sakon, H., & Tanaka, R. (2009). Paraprevotella clara gen. Nov., sp. Nov. And

Paraprevotella xylaniphila sp. Nov., members of the family “Prevotellaceae” isolated from human faeces. International Journal of Systematic and Evolutionary Microbiology, 59(Pt 8), 1895-1900. https://doi.Org/10.1099/ijs.0.008169-0

Parker, B. I, Wearsch, P. A., Veloo, A. C. M., & Rodriguez-Palacios, A. (2020). The Genus Alistipes: Gut Bacteria With Emerging Implications to Inflammation, Cancer, and Mental Health. Frontiers in Immunology, 11, 906. https://doi.org/10.3389/fimmu.2020.00906

Qin, I, Li, Y., Cai, Z., Li, S., Zhu, L, Zhang, F., Liang, S., Zhang, W., Guan, Y., Shen, D., Peng, Y., Zhang, D., Jie, Z., Wu, W., Qin, Y., Xue, W., Li, L, Han, L., Lu, D., ... Wang, J. (2012). A metagenome-wide association study of gut microbiota in type 2 diabetes. Nature, 490(7418), 55-60. https://doi.org/10.1038/naturell450 Rautio, M., Eerola, E., Vaisanen-Tunkelrott, M.-L., Molitoris, D., Lawson, P., Collins, M. D., & Jousimies-Somer, H. (2003). Reclassification of Bacteroides putredinis (Weinberg et al, 1937) in a new genus Alistipes gen. Nov., as Alistipes putredinis comb. Nov., and description of Alistipes finegoldii sp. Nov., from human sources. Systematic and Applied Microbiology, 26(2), 182-188. https://doi.org/10.1078/072320203322346029

Sakamoto, M., & Benno, Y. (2006). Reclassification of Bacteroides distasonis, Bacteroides goldsteinii and

Bacteroides merdae as Parabacteroides distasonis gen. Nov., comb. Nov., Parabacteroides goldsteinii comb. Nov. And Parabacteroides merdae comb. Nov. International Journal of Systematic and Evolutionary Microbiology, 56(Pt 7), 1599-1605. https://doi.Org/10.1099/ijs.0.64192-0 Samuel, B. S., & Gordon, J. I. (2006). A humanized gnotobiotic mouse model of host-archaeal-bacterial mutualism. Proceedings of the National Academy of Sciences of the United States of America, 103(26), 10011-10016. https://doi.org/10.1073/pnas.0602187103

Sircana, A., Framarin, L., Leone, N., Berrutti, M., Castellino, F., Parente, R., De Michieli, F., Paschetta, E., & Musso, G. (2018). Altered Gut Microbiota in Type 2 Diabetes: Just a Coincidence? Current Diabetes Reports, 78(10), 98. https://doi.org/10.1007/sll892-018-1057-6 Udayappan, S. D., Kovatcheva-Datchary, P., Bakker, G. J., Havik, S. R., Herrema, H., Cani, P. D., Bouter, K. E., Belzer, C., Witjes, J. J., Vrieze, A., de Sonnaville, E. S. V., Chaplin, A., vanRaalte, D. H., Aalvink, S., Dallinga-Thie, G. M., Heilig, H. G. H. J., Bergstrom, G., van der Meij, S., van Wagensveld, B. A., ... Nieuwdorp, M. (2017). Intestinal Ralstonia pickettii augments glucose intolerance in obesity. PloS One, 12(11), e0181693. https://doi.org/10.1371/joumal.pone.0181693 Woting, A., Pfeiffer, N., Loh, G., Klaus, S., & Blaut, M. (2014). Clostridium ramosum promotes high-fat diet- induced obesity in gnotobiotic mouse models. MBio, 5(5), e01530-01514. https://doi.org/10.1128/mBio.01530-14

Wu, H., Esteve, E., Tremaroli, V., Khan, M. T., Caesar, R., Mamieras-Holm, L., Stahlman, M., Olsson, L. M.,

Serino, M., Planas-Felix, M., Xifra, G., Mercader, J. M., Torrents, D., Burcelin, R., Ricart, W., Perkins, R., Fernandez-Real, J. M., & Backhed, F. (2017). Metformin alters the gut microbiome of individuals with treatment-naive type 2 diabetes, contributing to the therapeutic effects of the drag. Nature Medicine, 23(1), 850-858. https://doi.org/10.1038/nm.4345

Yamashita, H., FUJISAWA, K., 1TO, E., IDEI, S., KAWAGUCHI, N., K1MOTO, M., HIEMORI, M., & TSUJI, H. (2007). Improvement of Obesity and Glucose Tolerance by Acetate in Type 2 Diabetic Otsuka Long-Evans Tokushima Fatty (OLETF) Rats. Bioscience, Biotechnology, and Biochemistry, 71(5), 1236-1243. https://doi.org/10.1271/bbb.60668

[0094] While preferred embodiments of the present invention have been shown and described herein, it will be obvious to those skilled in the art that such embodiments are provided by way of example only. Numerous variations, changes, and substitutions will now occur to those skilled in the art without departing from the invention. It should be understood that various alternatives to the embodiments of the invention described herein may be employed in practicing the invention. It is intended that the following claims define the scope of the invention and that methods and structures within the scope of these claims and their equivalents be covered thereby.

Claims

CLAIMS WHAT IS CLAIMED IS:

1. A method for determining whether an individual is at risk for development or progression of glycemic dysregulation comprising a) providing a biological sample from the individual; b) sequencing nucleic acids from the sample to provide sequence information, wherein the nucleic acids comprise mRNA; c) determining, from the sequence information one or more features of the nucleic acid sequences; d) executing by computer a classification model that infers, from one or more features in the feature set, a measure of risk for development or progression of glycemic dysregulation in the individual.

2. The method of claim 1 wherein the one or more features comprise (1) determinations of KEGG-Orthologs (KOs) for gene-level activity within the sample.

3. The method of claim 1 or 2 wherein the determinations of KOs comprise at least 5, 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 85, 90, 95, 100, 110, 120, 130, 150, 170, or 200 different KOs and/or not more than 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 85, 90, 95, 100, 110, 120, 130, 150, 170, 200 or 300 KOs.

4. The method of any preceding claim wherein the one or more features comprise (2) measures of active microbial species within the sample, wherein the one or more measures are included in a feature set.

5. The method of any preceding claim wherein the measures of active microbial species include measures of species from at least 1, 2, 5, 7, 10, 12, 15, 17, 20, 22, 25, 30, 40, or 50 genuses and/or not more than 2, 5, 7, 10, 12, 15, 17, 20, 22, 25, 30, 40, 50 or 70 genuses.

6. The method of any preceding claim wherein the biological sample comprises a stool sample.

7. The method of any preceding claim wherein the individual is a human.

8. The method of any preceding claim wherein the individual has an HbAlc of less than 5.7% and is conventionally classified as nondiabetic.

9. The method of any preceding claim wherein the measure of risk for the individual indicates that the individual is at risk of developing glycemic dysregulation

10. The method of any preceding claim wherein the individual has an HbAlc of 5.7-6.4% and is conventionally classified as prediabetic.

11. The method of any preceding claim wherein the measure of risk for the individual indicates that the individual is at risk of progressing glycemic dysregulation.

12. The method of any preceding claim wherein the individual has an HbAlc of 6.5% or greater and is conventionally classified as diabetic.

13. The method of any preceding claim wherein the measure of risk for the individual indicates that the individual is at risk of progressing glycemic dysregulation

14. The method of any preceding claim wherein the sequence information comprises metatranscriptomic information.

15. The method of any preceding claim wherein the feature set used by the classification algorithm includes at least measures of activity of one or more microbial taxa.

16. The method of any preceding claim wherein the feature set used by the classification algorithm includes at least measures of taxa involved in one or more metabolic pathways.

17. The method of any preceding claim wherein the classification model uses expression levels for one or more species of one or more genus shown in Figure 9a.

18. the method of any preceding claim wherein the classification model uses expression levels for species of at least, exactly, and/or no more than any of 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30 of the genuses shown in Figure 9a.

19. The method of any preceding claim wherein a higher risk of developing or progressing glycemic dysregulation is determined, at least in part, from relatively lower expression of Ruminococus species compared to nondiabetic individuals.

20. The method of claim 19 wherein the Ruminococus species comprise Ruminococcus bicirculans , R. callidus, and/or R. champallensis .

21. The method of any preceding claim wherein a higher risk of developing or progressing glycemic dysregulation is determined, at least in part, from relatively lower expression of Blautia species in the sample compared to nondiabetic controls.

22. The method of claim 21 wherein the Blautia species comprise Blautia massiliensis and/or Blautia Marseille-P3087.

23. The method of any preceding claim wherein the classification model uses at least, exactly, and/or no more than any of 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21,

22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73

74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, or 95 of the KOs selected from those of Figure 9b.

24. The method of any preceding claim further comprising outputting the inference to a user interface device or to a computer-readable memory.

25. A method comprising delivering and/or administering to an individual one or more therapeutic interventions to decrease glycemic dysregulation or the development of glycemic dysregulation, wherein the individual has been found to be at risk for developing or progressing glycemic dysregulation by the method of any preceding claim.

26. The method of claim 25 wherein the therapeutic intervention comprises administering one or more pharmaceutical agents to the individual.

27. The method of claim 25 or 26 wherein the one or more pharmaceutical agents comprise metformin, a sodium glucose co-tranporter 2 inhibitor (SGLT2i), a glucagon-like peptide 1 receptor agonist (GLP-RA), or a combination thereof.

28. The method of claim 27 wherein the pharmaceutical agent comprises metformin.

29. The method of claim 28 wherein 500 mg of metformin is taken daily.

30. The method of claim 28wherein 1000 mg of metformin is taken daily.

31. The method of claim 28 wherein 1500 mg of metformin is taken daily.

32. The method of claim 28 wherein 2000 mg of metformin is taken daily.

33. The method of claim 28 wherein 2500 mg of metformin is taken daily.

34. The method of any one of claims 26 through 33 wherein the pharmaceutical agent comprises a SGLT2i selected from the group consisting of canagliflozin, dapagliflozin, and empagliflozin

35. The method of any one of claims 26 through 34 wherein the pharmaceutical agent comprises a GLP-RA selected from the group consisting of Dulaglutide, liraglutide, lixisenatide, Exenatide, Exenatide extended release, Semaglutide, injection, and Semaglutide, oral.

36. The method of any one of claims 25 through 35 wherein the therapy comprises a dietary therapy.

37. The method of claim 36 wherein the dietary therapy comprises eating a whole foods plant- based diet and reducing or eliminating meat; reducing or eliminating cheese consumption, eating the largest meal of the day at midday, or a combination thereof.

38. The method of claim 36 or claim 37 wherein the dietary therapy comprises eating a diet of an average of less than 200, 150, 100, or 50 grams of carbohydrate daily, periodically fasting for at least 16 hours, or a combination thereof.

39. The method of any one of claims 25 through 38 wherein the therapy comprises exercise.

40. The method of any one of claims 25 through 39 wherein the therapy comprises administration of one or more supplements.

41. The method of claim 40 wherein the supplements comprise one or more pre-biotics, one or more probiotics, or a combination thereof.

42. The method claim 41 wherein the one or more prebiotics and/or probiotics are chosen to modulate the gut microbiome of the individual toward a biome that favors decreased glucose dysregulation.

43. A method for determining whether an individual is able to control blood glucose with metformin comprising a) providing a biological sample from the individual; b) sequencing nucleic acids from the sample to provide sequence information, wherein the nucleic acids comprise mRNA; c) determining, from the sequence information one or more features of the nucleic acid sequences; d) executing by computer a classification model that infers, from one or more features in the feature set, a measure of likelihood of control of blood glucose with metformin for the individual.

44. The method of claim 43 wherein the one or more features comprise (1) determinations of KEGG-Orthologs (KOs) for gene-level activity within the sample.

45. The method of claim 43 or 44 wherein the determinations of KOs comprise at least 5, 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 85, 90, 95, 100, 110, 120, 130, 150, 170, or 200 different KOs and/or not more than 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 85, 90, 95, 100, 110, 120, 130, 150, 170, 200 or 300 KOs.

46. The method of any of claims 43-45 wherein the one or more features comprise (2) measures of active microbial species within the sample, wherein the one or more measures are included in a feature set.

47. The method of any of claims 43-46 wherein the measures of active microbial species include measures of species from at least 1, 2, 5, 7, 10, 12, 15, 17, 20, 22, 25, 30, 40, or 50 genuses and/or not more than 2, 5, 7, 10, 12, 15, 17, 20, 22, 25, 30, 40, 50 or 70 genuses.

48. The method of any of claims 43-47 wherein the biological sample comprises a stool sample.

49. The method of any of claims 43-48 wherein the individual is a human.

50. The method of any of claims 43-49 wherein the individual has an HbAlc of less than 5.7% and is conventionally classified as nondiabetic.

51. The method of any of claims 43-50 wherein the individual has an HbAlc of 5.7-6.4% and is conventionally classified as prediabetic.

52. The method of any of claims 43-51 wherein the individual has an HbAlc of 6.5% or greater and is conventionally classified as diabetic.

53. The method of any of claims 43-52 wherein the sequence information comprises metatranscriptomic information.

54. The method of any of claims 43-53 wherein the feature set used by the classification algorithm includes at least measures of activity of one or more microbial taxa.

55. The method of any of claims 43-54 wherein the feature set used by the classification algorithm includes at least measures of taxa involved in one or more metabolic pathways.

56. The method of any of claims 43-55 wherein the classification model uses expression levels for one or more species of one or more genuses shown in Figure 10c.

57. The method of any of claims 43-56 wherein the classification model uses expression levels for species of at least, exactly, and/or no more than any of 1, 2, 3, 4, or 5, of the genuses shown in Figure 10c.

58. The method of any of claims 43-57 wherein the classification model uses at least, exactly, and/or no more than any of 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21,

74, 75, 76, 77, 78, 79, 80, or 81, of the KOs selected from those of Figure lOd.

59. The method of any of claims 43-58 wherein 500 mg of metformin is taken daily.

60. The method of any of claims 43-59 wherein 1000 mg of metformin is taken daily.

61. The method of any of claims 43-60 wherein 1500 mg of metformin is taken daily.

62. The method of any of claims 43-61 wherein 2000 mg of metformin is taken daily.

63. The method of any of claims 43-62 wherein 2500 mg of metformin is taken daily.

64. A method comprising: a) providing biological samples from each of a first set of subjects, a second set of subjects, and a third set of subjects, wherein the biological samples comprise a stool samples, and wherein the first set of subjects have been diagnosed with Type 2 diabetes, the second set of subjects have been diagnosed as prediabetic, and the third set of subjects have been diagnosed as nondiabetic; b) sequencing nucleic acids in the biological samples to provide sequence information; and c) performing a statistical analysis on the sequence information to produce a model that predicts a risk of developing or progressing glucose dysregulation in an individual.

65. The method of claim 64 wherein the statistical analysis comprises an analysis selected from correlational, Pearson correlation, Spearman correlation, chi-square, comparison of means (e.g., paired T-test, independent T-test, ANOVA) regression analysis (e.g., simple regression, multiple regression, linear regression, non-linear regression, logistic regression, polynomial regression stepwise regression, ridge regression, lasso regression, elasticnet regression) and non-parametric analysis (e.g., Wilcoxon rank-sum test, Wilcoxon sign-rank test, sign test).

66. A method comprising administering a therapy to normalize or improve blood glucose regulation to an individual found to be at risk of developing or progressing blood glucose dysregulation by the method of claim 1.

67. The method of claim 66 wherein the therapy comprises any of the therapies described for claim 1.

68. A system comprising:

(i) a computer comprising:

(a) a processor; and

(b) a memory storage module, coupled to the processor, the memory storing a module comprising:

(1) nucleic acid sequence information from a biological sample from a subject comprising a gut microbiome; (2) a classification model which, based on values including the measurements, classifies the subject as at risk for developing or progressing glucose dysregulation, wherein the classification model is selected to have a sensitivity of at least 75%, at least 85% or at least 95%; and

(3) computer executable instructions for implementing the classification model on the test data.

69. A method for developing a computer model for inferring, from feature data, a risk of developing or progressing glucose dysregulation in a subject comprising: a) training a machine learning algorithm on a training data set, wherein the training data set comprises, for each of a plurality of subjects, (1) a class label classifying a subject as nondiabetic, prediabetic, or diabetic; and (2) feature data comprising quantitative measures for each of a plurality of features selected from gut microbiome transcriptome expression, and wherein the machine learning algorithm develops a model that infers a class label for a subject based on the feature data.

70. A method that infers a risk of developing or progressing glucose dysregulation in a subject, the method comprising:

(a) providing a data set comprising, for the subject, feature data for each of a plurality of features selected from gut microbiome transcriptome gene expression data and taxa activity data; and

(b) executing a computer model on the data set to infer the risk of developing or progressing glucose dysregulation in the subject.

71. A software product comprising a computer readable medium in tangible form comprising machine executable code, which, when executed by a computer processor, infers a risk of developing or progressing glucose dysregulation in a subject by:

(a) accessing a data set comprising, for a subject, feature data for each of a plurality of features selected from gut microbiome transcriptome gene expression data and taxa activity data; and

72. A method of providing therapy to decrease a risk of developing or progressing glucose dysregulation or to normalize glucose regulation in a subject comprising: (a) inferring the presence of a risk of developing or progressing glucose dysregulation in a subject according to a method as described herein; and

(b) administering a therapeutic intervention to the subject effective to decrease the risk of developing or progressing glucose dysregulation or to normalize glucose regulation in the subject.

73. A method for diagnosing and treating developing or progressing glucose dysregulation in a subject, the method comprising:

(a) receiving from a subject a sample comprising a gut microbiome;

(b) determining nucleic acid sequences of a microorganism component of the sample;

(c) determining alignments of the nucleic acid sequence to reference nucleic acid sequences associated with developing or progressing glucose dysregulation;

(d) generating a microbiome feature dataset for the subject based upon the alignments;

(e) generating an inference of the risk of developing or progressing glucose dysregulation in the subject upon processing the microbiome feature dataset with an inference model derived from a population of subjects; and

(1) at an output device associated with the subject, providing a therapy to the subject with the risk of developing or progressing glucose dysregulation upon processing the inference with a therapy model designed to treat the glucose dysregulation.

74. A method comprising:

(a) measuring, in a sample from a subject comprising an gut microbiome, activity of one or more biomarkers selected from Figure 9;

(b) inferring, from the measurements, a risk of developing or progressing glucose dysregulation in the subject; and

(c) delivering to the subject a therapeutic intervention effective to decrease the risk of developing or progressing glucose dysregulation or to normalize glucose regulation in the subject.

75. A method comprising: a) providing biological samples from each of a first set of subjects and a second set of subjects having glucose dysregulation and having been subject to a therapeutic intervention, wherein the biological samples comprise a gut microbiome and wherein the first set of subjects responded positively to the therapeutic intervention and the second set of subjects did not respond positively to the therapeutic intervention; b) sequencing nucleic acids in the biological samples to provide sequence information; and c) performing a statistical analysis on the sequence information to produce a model that infers having a positive response or lack of positive response to the therapeutic intervention in a subject.

76. A method of treating a subject with glucose dysregulation comprising:

(a) inferring that the subject will respond positively to each of one or more therapeutic interventions by executing a model on nucleic acid information from a biological sample from the subject comprising a gut microbiome; and

(b) administering to the subject one or more therapeutic interventions to treat the glucose dysregulation

77. The method of claim 76 wherein the therapeutic intervention comprises administration of metformin.

78. A method comprising at a computer system comprising at least one processor and a memory, storing at least one program for execution by the at least one processor: a) obtaining nucleic acid sequence data in electronic form for a first plurality of nucleic acids from a biological sample of the subject, wherein the nucleic acid sequence data comprises sequences for 500, 1000, 2000, 5000, 7000, or 10,000 or more nucleic acids, preferably 1000 or more, more preferably 5000 or more, even more preferably 10,000 or more nucleic acids, in the plurality of nucleic acids; b) obtaining phenotypic data about the subject in electronic form, the phenotypic data comprising a plurality of responses; c) identifying, from the nucleic acid sequence data, a corresponding presence or absence of each respective functional activity condition in a plurality of functional activity conditions in the subject by:

1) determining a corresponding measure of transcriptional activity for each respective KEGG ortholog designation in a plurality of KEGG ortholog designations by matching each respective nucleic acid in the first plurality of nucleic acids to individual sequences of the plurality of KEGG ortholog designations,

2) determining a corresponding measure of transcriptional activity for each respective microbial taxa in a plurality of microbial taxa by matching each respective nucleic acid in the first plurality of nucleic acids to individual sequences of the plurality of taxa, 3) determining, for each respective functional activity category in the plurality of functional activity categories, a corresponding functional activity score based on a corresponding measure of transcriptional activity for a respective KEGG ortholog designation in the plurality of KEGG ortholog designations or a corresponding measure of transcriptional activity for a respective microbial taxa in the plurality of microbial taxa, and

4) identifying, for each respective functional activity condition in the plurality of functional activity conditions, the corresponding presence or absence of the respective functional activity condition based on a corresponding functional activity score; d) identifying, from the phenotypic data from the subject, a corresponding presence or absence of, or degree of, each respective phenotypic condition in a plurality of phenotypic conditions in the subject by:

1) assigning a corresponding numerical value to each respective response in the plurality of responses comprising the phenotypic data,

2) determining, for each respective phenotypic condition in the plurality of phenotypic conditions, a corresponding phenotype score based on a corresponding numerical value for one or more respective response in the plurality of responses, and

3) identifying, for each respective phenotypic condition in the plurality of phenotypic conditions, the corresponding presence or absence of, or degree of, each respective phenotypic condition in the plurality of phenotypic conditions based on a corresponding phenotypic score; e) accessing a knowledge base that includes for each of a plurality of food items (i) a corresponding first desirability ranking, in a plurality of desirability rankings, of the food for each respective phenotypic condition in the plurality of phenotypic conditions present in the subject and (ii) a corresponding second desirability ranking, in the plurality of desirability rankings, of the food for each respective functional activity condition in the plurality of functional activity conditions present in the subject;

1) using a recommendation engine, executing logic to produce a corresponding recommendation for each respective food item in the plurality of food items for the subject based on the corresponding first desirability rankings and the corresponding second desirability rankings for the respective food item, thereby generating a plurality of food recommendations; and g) outputting the plurality of food recommendations to an electronic device accessible by the subject.