KR20190032847A - miRNA and mRNA ASSOCIATION ANALYSIS METHOD AND GENERATING APPARATUS FOR miRNA and mRNA ASSOCIATION NETWORK - Google Patents

Abstract

The miRNA-mRNA association assay method comprises the steps of: generating a sample data including a plurality of miRNA expression data and expression data of a plurality of mRNAs, wherein the computer device is operable to detect at least one of the plurality of miRNAs and the plurality of mRNAs Generating a plurality of first association models for at least one of the plurality of first association models and the computer device combining the plurality of first association models to generate a second association model.

Description

miRNA-mRNA association analysis method and miRNA-mRNA network generation apparatus <br> <br> <br> Patents - stay tuned to the technology miRNA-mRNA association analysis method and miRNA-

The technique described below relates to a technique for analyzing the relationship between miRNA and mRNA.

mRNA (messenger RNA), mRNA, is transcribed from a DNA prototype and carries it to the ribosomes where the protein synthesis takes place by giving a codon to the genetic information. mRNA is directly involved in protein synthesis by transmitting information for protein synthesis. MicroRNAs (miRNAs) are small non-expressed RNA molecules that are found in plants, animals, viruses, and are composed of about 22 nucleotides. miRNAs inhibit (control) the translation process for specific mRNAs. In other words, miRNA does not provide codon information for protein synthesis, but it functions to control protein synthesis constantly. miRNAs are known to be involved extensively in biological processes such as cell differentiation, intercellular signaling or specific tumors. miRNA plays an important role in the process of controlling gene expression. Therefore, we have a functional relationship between a specific miRNA and a specific mRNA. Studies are underway on the relationship between miRNA and mRNA.

miRNA binds to complementary sequences in the 3'-untranslated regions of mRNA to regulate post-transcription to specific mRNAs. Based on these properties, there was a correlation between miRNA and mRNA using sequence data based on the complementarity or structural stability of the sequence.

In order for a particular miRNA to control a particular mRNA, the level of expression of that miRNA may be altered. That is, there is a certain correlation between miRNA expression and mRNA expression. Based on these properties, there has been research into network construction to confirm the association between miRNA and mRNA.

Joung JG, Hwang KB, Nam JW, Kim SJ, Zhang BT. Discovery of microRNA-mRNA modules via population-based probabilistic learning. Bioinformatics. 2007; 23: 1141-7p.

(1) Sequence data-based analytical methods may be suitable for analyzing the association between specific miRNAs and mRNAs, but have high false positives and false negatives. (2) The analysis method using the correlation between the expression of miRNA and mRNA (i) It is difficult to exclude that miRNA indirectly affects mRNA by using gene expression data. In other words, there is a high probability that there is a false edge in the network of miRNA and mRNA. (ii) Microarray data is also basically extracted from a limited sample. Therefore, there is a high-dimensional low-sample size problem in the relationship between miRNA and mRNA. (iii) Further, it may be difficult to predict in advance whether an analysis technique is more effective for specific expression data for miRNA-mRNA association analysis. For example, if one technique is applied in various experimental environments, the experimental results may be inconsistent.

The techniques described below attempt to solve the conventional problems and provide a technique for analyzing the relationship between miRNA and mRNA.

The miRNA-mRNA association assay method comprises the steps of: generating a sample data including a plurality of miRNA expression data and expression data of a plurality of mRNAs, wherein the computer device is operable to detect at least one of the plurality of miRNAs and the plurality of mRNAs Generating a plurality of first association models for at least one of the plurality of first association models and the computer device combining the plurality of first association models to generate a second association model.

The miRNA-mRNA network generating apparatus includes an input device for receiving expression data of a plurality of miRNAs and expression data of a plurality of mRNAs, a storage device for storing a program for building a miRNA-mRNA network, the method comprising: executing the program using mRNA expression data as input data to generate sample data including input data and expression data of the plurality of miRNAs and expression data of a plurality of mRNAs; generating a plurality of first association models for at least one of the miRNAs and the plurality of mRNAs, integrating the plurality of first association models to generate a second association model, A miRNA-m &lt; / RTI &gt; indicating a degree of association of at least one of the plurality of miRNAs with at least one of the plurality of mRNAs And an arithmetic unit for generating an RNA network.

The techniques described below provide a miRNA-mRNA network that reliably correlates miRNAs with mRNA even when using small sample data.

Figure 1 is an example of a process for constructing a miRNA-mRNA network.
Figure 2 is an example of a miRNA-mRNA network.
Figure 3 is an example of a procedure flow diagram for miRNA-mRNA network construction.
4 is an example of a miRNA-mRNA network construction apparatus.
FIG. 5 shows an example of an experimental result that verifies the effect of the proposed technique.

The following description is intended to illustrate and describe specific embodiments in the drawings, since various changes may be made and the embodiments may have various embodiments. However, it should be understood that the following description does not limit the specific embodiments, but includes all changes, equivalents, and alternatives falling within the spirit and scope of the following description.

The terms first, second, A, B, etc., may be used to describe various components, but the components are not limited by the terms, but may be used to distinguish one component from another . For example, without departing from the scope of the following description, the first component may be referred to as a second component, and similarly, the second component may also be referred to as a first component. And / or &lt; / RTI &gt; includes any combination of a plurality of related listed items or any of a plurality of related listed items.

As used herein, the singular " include " should be understood to include a plurality of representations unless the context clearly dictates otherwise, and the terms &quot; comprises & , Parts or combinations thereof, and does not preclude the presence or addition of one or more other features, integers, steps, components, components, or combinations thereof.

Before describing the drawings in detail, it is to be clarified that the division of constituent parts in this specification is merely a division by main functions of each constituent part. That is, two or more constituent parts to be described below may be combined into one constituent part, or one constituent part may be divided into two or more functions according to functions that are more subdivided. In addition, each of the constituent units described below may additionally perform some or all of the functions of other constituent units in addition to the main functions of the constituent units themselves, and that some of the main functions, And may be carried out in a dedicated manner.

Also, in performing a method or an operation method, each of the processes constituting the method may take place differently from the stated order unless clearly specified in the context. That is, each process may occur in the same order as described, may be performed substantially concurrently, or may be performed in the opposite order.

The technique described below relates to a technique for analyzing the relationship between microRNA (miRNA) and mRNA. The technique described below analyzes the relationship between miRNA and mRNA to be analyzed using microarray data.

The technique described below establishes a miRNA-mRNA network that shows the association of miRNA and mRNA from gene expression data. The miRNA-mRNA network refers to a graphical data structure to show the relationship between miRNA and mRNA.

Figure 1 is an example of a process for constructing a miRNA-mRNA network. First, miRNA expression data and mRNA expression data are obtained (A). The gene expression data may be, for example, the result of a microarray experiment. The gene expression data may be output data or image data on which the microarray experiment is performed as shown in FIG. Researchers can derive the expression status of a specific gene using gene expression data. Or the computer device can automatically read and analyze the microarray experiment result data to derive the expression state of a specific gene. Since the computer device constructs the miRNA-mRNA network, it is assumed that the gene expression data is converted into digital data. Then, the computer device derives a model (matrix) representing the correlation of miRNA and mRNA (B). Finally, a computer device constructs a miRNA-mRNA network using a model that shows the correlation of miRNA and mRNA. The detailed process will be described later.

Figure 2 is an example of a miRNA-mRNA network. Figure 2 corresponds to the result produced in Figure 1 (C). In Figure 2, the circle represents mRNA and the square represents miRNA. When the nodes (miRNA and mRNA) are connected to the edge in the resulting graph, it means that the miRNA and mRNA are directly related to each other. In the miRNA-mRNA network of FIG. 2, the edge-linked miRNA and mRNA are directly related. For example, miRNA a is associated with four mRNAs 1, 2, 3 and 4. In other words, miRNA a is involved in the control mechanisms of mRNAs 1, 2, 3 and 4. mRNA 4 is associated with three miRNAs a, b, and d. That is, miRNAs a, b, and d are involved in the control mechanism of mRNA 4.

The techniques described below use several techniques to construct a miRNA-mRNA network with direct association. Briefly, (1) techniques for identifying direct associations, (2) bootstrapping techniques, and (3) ensemble techniques are used. First, we exclude indirect associations predicted from expression data. At least one of (1) Partial correlation estimation, (2) Sparse Partial Correlation Estimation (SPACE), and (3) Network deconvolution may be used to identify only direct associations. This method is intended to infer the direct correlation between miRNA and mRNA. Second, in order to solve the problem of high-dimension low-sample size problem, the experiment identifies miRNAs and mRNAs expressed in different aspects to reduce the size of the data set. We also use bootstrapping techniques to maximize the limited sample size. Third, non-parametric Ensemble techniques are used to increase the reliability of the relationship between miRNA and mRNA inferred. A rank-based non-parametric Ensemble is used to integrate the bootstrapping technique and the different direct association identification results.

Figure 3 is an example of a procedure flow diagram for miRNA-mRNA network construction. Hereinafter, it is assumed that the miRNA-mRNA network construction process is performed on a computer device.

First, miRNA and mRNA expression data are required to analyze the relationship between miRNA and mRNA. It is assumed that the target of the analysis is the expression data of matched miRNA and mRNA. The computer device receives the expression data of the miRNA and the expression data of the mRNA (110, 120). The computer device is presumed to be able to confirm the expression status of a specific gene using expression data. On the other hand, the computer device can constantly pre-process the input expression data. Pretreatment is the process of normalizing expression data for later analysis.

The computer device identifies the differentially expressed genes in miRNA expression data and mRNA expression data to reduce the dimensionality of the data (115, 125). The differentially expressed gene means a gene whose expression is significantly increased or decreased in the test sample as compared with the control sample. Generally, differentially expressed genes are identified by comparing them with the average values in microarray data. For example, a SAM (Significant Anlysis of Microarray) method or the like can be used.

This is to focus on the association of active miRNA with mRNA. Since miRNA expression data and mRNA expression data are obtained on different platforms, the selected expression data must be integrated. The computer device integrates the selected data of miRNA and mRNA expression data (130). In FIG. 3, the highest number in the table representing the expression data corresponds to the identifier of the gene. The computer device may also perform scaling of miRNA expression data and mRNA expression data.

Further, the computer device performs bootstrapping (140) on the integrated miRNA and mRNA expression data to solve the small sample size problem. Bootstrapping generates m new training data sets. Bootstrapping is a technique for generating various versions of a model. Bootstrapping is a well-known technique, so a detailed description is omitted. Bootstrapping repeatedly extracts sample data from a population sample to generate a new data set. The data contained in the new data set is composed of values in the original sample data. Performing m-times bootstrapping on the integrated miRNA and mRNA expression data described above produces m new data sets.

The computer device applies three direct correlation inference techniques (150, 160 and 170) to the integrated miRNA expression data and mRNA expression data to derive the direct miRNA-to-mRNA association. Inference techniques to derive direct associations generate direct correlation models from phenotypic data. The correlation model has a matrix form containing all possible combinations between miRNA and mRNA. The computer device derives a direct correlation using m samples from the bootstrapping results for each of the three direct correlation inference techniques. Each correlation inference method generates m models using m samples. In the computer device, m models generated by each correlation inference technique generate a single model using rank-based aggregation (155, 165, and 175). FIG. 3 shows an example in which all three techniques are used. In some cases, however, at least one of the three techniques described in FIG. 3 (partial correlation coefficient estimation, lean partial correlation estimation and network deconvolution) may be used.

In order to integrate models generated by different correlation inference techniques, computer devices use rank-based ensemble techniques (180). Whereby the computer device eventually generates one direct correlation model.

A direct miRNA-mRNA network can be constructed by applying a constant threshold to the value of the final direct correlation model (matrix) (190). For example, if the threshold is 50, then the computer device will filter (ignore) less than 50 in the direct correlation model and determine that the matrix with values greater than 50 has a direct correlation. A computer device can construct an miRNA-mRNA network by linking edges to miRNA and mRNA pairs with a matrix value of 50 or greater.

The three direct correlation inference methods are explained in detail. The direct correlation inference technique is intended to exclude the indirect correlation that may appear in the miRNA-mRNA network as described above.

① Partial correlation estimation method

The partial correlation estimator measures the association weights between two random variables by suppressing the effect of a set of random variables. The method based on partial correlation can infer conditional dependency through nonzero entries in the concentration matrix, which is an inverse of the covariance matrix. Analysis of the gene network using the partial correlation estimation method indicates that the zero item has no direct effect between the two nodes. That is, a nonzero entry means that there is a direct association between the two nodes. (See, for example, Schafer J, Strimmer K. A shrinkage approach to large-scale covariance matrix estimation and implications for functional genomics. Statist Appl Genet Mol Biol. 2005; 4: 32 ). The partial correlation estimator can be used to estimate one or more miRNAs and mRNAs that are directly related to miRNA expression data and mRNA expression data.

② Lean partial correlation estimation method (SPACE)

SPACE is another technique for calculating partial correlation coefficients in the presence of small samples (Peng J, Wang P, Zhou N, Zhu J. J Am Stat Assoc - Theory and Methods. 2009; 104 (486): 735-46). SPACE assumes that the matrix representing the partial correlation is sparse and that most pairs of variables have conditional independence. SPACE yields a sparse matrix whose elements are mostly zero. SPACE eventually helps to estimate the nonzero correlation. SPACE uses sparse regression techniques to estimate sparse partial correlation.

(3) Network deconvolution

The network deconvolution technique is a technique for inferring a direct dependency network by removing the indirect weight of the dependency network from the input data (Feizi S, Marbach D, Medard M, Kellis M. Network deconvolution as a general method to distinguish direct dependencies in networks Nat Biotechnol. 2013; 31: 726-33). The network deconvolution technique assumes that the relevance weights measured from the input data are the sum of the direct and indirect weights. Furthermore, the network deconvolution technique assumes that the indirect information flow can be estimated directly from the association weights.

Let G _obs be the observed independence network, G _tru be the true independence network, and G _ind be the indirect independence network. G _obs can be expressed as Equation 1 below.

Therefore, the network deconvolution technique derives a true direct dependency network by reversing the effect of passing information across all possible indirect paths. As a result, the true direct dependency network is given by Equation 2 below.

The computer device estimates the observed correlation between miRNA and mRNA expression data using reciprocal information. The computer device then applies deconvolution techniques to the correlation model between miRNA and mRNA to exclude indirect correlation.

Each of the three direct correlation inference techniques described above generates one model (represented by a matrix). We now describe procedures for integrating miRNA-mRNA-correlated models with different techniques. The computer device incorporates different models using the rank-based ensemble technique described above. It is assumed that the computer system uses the rank - based nonparametric ensemble technique.

Traditional rank-based integration techniques use a rank-sum-based approach, an average-rank-based approach, or a Borda count election. In order to integrate the results of the three direct correlation inference techniques described below, a reverse rank based method is used.

Each matrix generated by the direct association inference technique represents a direct correlation to the total miRNA-mRNA combination, which is the weight value of the edge on the miRNA-mRNA-related network. In order to integrate the results obtained by the direct correlation inference method non-parametrically, the associative numerical values of the result matrices are converted into the descending order, and the reverse rank value is used as the relevance weight in order to assign a high weight to the small . On the other hand, the weight between a specific miRNA and mRNA in the integrated network is calculated as the product of the reverse rankings of the results of all direct association inference techniques of the corresponding miRNA-mRNA pair. That is, the edge between a specific miRNA and mRNA is determined by multiplying the reverse rank of the same edge among the three direct association inference results. For example, if all three inference results include the same edge, the value obtained by multiplying all three reverse rankings is the weight of the edge. Or if two of the three inference results (the model) contain the same edge, then the product of the two reverse ranks is the weight of that edge.

Let r _ij be the associative weight set of nodes i and j. The relevance weight r ' _ij of the integrated graph using the rank-based strategy can be expressed as Equation 3 below. Equation (3) below reduces the computational complexity by converting the multiplication operation to the hypothesis by log operation.

On the other hand, it is also possible to use the reverse ranking approach when merging the m results generated by one direct associative analogy.

In the foregoing description, it has been stated that the computer device constructs the miRNA-mRNA network. Hereinafter, a computer apparatus for constructing a miRNA-mRNA network will be described. 4 is an example of a miRNA-mRNA network construction apparatus. The construction apparatus described in Fig. 4 corresponds to the computer apparatus described above.

16A is an example of an apparatus 200 for constructing a miRNA-mRNA network by receiving gene expression data from an object located at a network end. The miRNA-mRNA network building device 200 includes a gene expression DB 210 and a computer device 220.

The gene expression DB (expression DB, 210) stores data related to gene expression of a specific organism. As described above, the gene expression data is prepared using a technique such as a microarray. Gene expression data includes miRNA expression data and mRNA expression data.

The computer device 220 receives the expression data stored in the gene expression DB 210. As described above, the computer device 220 applies statistical techniques to the gene expression data to select the differentially expressed genes and integrates the selected gene expression data. The computer device 220 increases the sample of the consolidated data through bootstrapping. The computer device 220 generates correlation models by applying three techniques for direct association analysis based on the sample data. The computer device 220 integrates the results of the three techniques to produce a final correlation model. Finally, the computer device 220 constructs the miRNA-mRNA network based on the final correlation model. Each process for constructing the miRNA-mRNA network is as described above.

Fig. 4 (B) is an example of a computer device 300 for building a miRNA-mRNA network. The computer device 300 for constructing the miRNA-mRNA network includes an input device 310, a computing device 320, a storage device 330, and an output device 340.

The input device 310 receives the gene expression data. The input device 310 may be a physical interface device such as a keyboard, a mouse, and a touch pad. Alternatively, the input device 310 may be a device that receives the stored gene expression data from an external storage medium (USB, etc.). Or the input device 310 may be a communication module that receives gene expression data from an external network.

The storage device 330 stores a program for building the miRNA-mRNA network. The storage device 330 may also store the constructed miRNA-mRNA network.

 The computing device 320 performs an operation for constructing the miRNA-mRNA network using the input gene expression data and the program stored in the storage device 330. The process of constructing the miRNA-mRNA network is the same as that performed by the computer device 220 described above. Further, the computing device 320 may perform a process of analyzing the miRNA-mRNA network constructed using the analysis program stored in the storage device 330.

The output device 340 is a device that outputs the miRNA-mRNA network or analysis result. The output device 340 may be a display device for outputting images, a printer for outputting text, and the like. Furthermore, the output device 340 may be a communication module that transmits the generated miRNA-mRNA network or analysis data to another device.

Hereinafter, the effect on the technique of analyzing miRNA-mRNA association described above will be described. Hereinafter, the technique for analyzing the miRNA-mRNA association is referred to as the proposed technique. To verify the effectiveness of the proposed method, we use widely known miRNA and mRNA expression data. First, we examine the effect of the bootstrapping and ensemble technique, which is used in conjunction with the technique of estimating the direct association between miRNA and mRAN in the proposed scheme, and compare the effectiveness of the proposed technique with the ensemble technique known to have the best performance at present.

Three matched miRNAs and mRNA expression data sets used in previous studies were used (Le TD, Zhang J, Liu L, Li J. Ensemble methods for miRNA target prediction expression data. PLoS One. ): e0131627 and Le TD, Zhang J, Liu L, Liu H, Li J. MiRLAB: an R based dry lab for exploring miRNA-mRNA regulatory relationships. The three expression data sets are EMT (Epithelial to Mesenchymal Transition) data, MCC (Multi-Class Cancer) data, and BR (Breast Cancer) data. EMT data is the matched miRNA-mRNA expression data for 11 samples of epithelia and 36 samples of mesenchymal. MCC data are matched miRNA-mRNA expression data obtained from normal tissues and tumor tissues of eight organs in 60 samples. BR data is the matched miRNA-mRNA expression data obtained in 50 samples of breast cancer tissues.

A commercial program can be used to screen differential expression data. The p value is used as a criterion for distinguishing differentiation expression. 35 miRNAs and 1,154 mRNAs were selected from the EMT data by limiting the p value to less than 0.05, and 108 miRNAs and 1,860 mRNAs were selected from the MCC data. In the BR data, 92 miRNAs were selected by limiting the p value to less than 0.2, and 1,500 mRNAs were selected by limiting the p value to less than 0.0001. Selected differential expression data were integrated and regularly preprocessed (normalized) for further processing.

The correlation between miRNA and mRNA has been extrapolated to the present experimental results. Therefore, it is assumed that the correlation according to the results of the conventional research is a true correlation, and the effect of the proposed technology is verified. Conventional experimental results (true correlation criterion) are based on four databases Tarbase v.6.0 (Vergoulis T, Vlachos IS, Alexiou P, Georgakilas G, Maragkakis M, Reczko M, Tarbase 6.0: capturing the exponential growth of miRecords v2013 (Xiao F, Zuo Z, Cai G, Kang S, Gao X, and Li T. miRecords: an integrated resource for miRNA targets with experimental support. Nucleic Acids Res. 2012; 40 (D1): D222-9. microRNA target interactions. Nucleic Acids Res. 2009; 37 suppl 1: D105-10.), miRWalk v2.0 (Dweep H, Sticht C, Pandey P, Gretz N. miRWalk database: prediction of possible miRNA binding sites by walking the genes of three genomes. J Biomed Inform. 2011; 44 (5): 839-47) and miRTarBase v.4.5 (Hsu SD, Tseng YT, Shrestha S, Lin YL, Khaleel A, Chou CH, et al. : an information resource for experimentally validated miRNA target interactions. Nucleic Acids Res. 2014; 42 (D1): D78-85.) were used as validation data. The accuracy of the proposed method was evaluated based on the number of intersection elements between the collection of k mRNA pairs ranked high for each miRNA and the verification data in all miRNA and mRNA pairs estimated for association.

FIG. 5 shows an example of an experimental result that verifies the effect of the proposed technique. FIG. 5 (A) shows an example of the effect of various embodiments proposed in the proposed method for estimating the direct association of miRNA with mRNA. The effects of three gene expression data (EMT, MCC and BR) were verified. For each miRNA, the effect was analyzed based on the top 100 miRNA-mRNA pairs.

The sample data is divided into the case where the entire expression data is used as it is (Whole) and the case where the sample data is amplified through bootstrapping (Bootstrap). A single method is one that uses one technique for direct correlation inference. The three techniques are partial correlation coefficient estimation (Corpcor (C)), SPACE (S) and network deconvolution (ND (N)). The ensemble method is the combination of C and S (C & S), C and N (C & N), S and N (S & N) .

First, when a single technique is applied to whole expression data (Whole), it can be seen that different techniques are effective depending on the types of gene expression data. For example, EMT was the most effective, S was the MCC, N was the MCC, and C was the BR. That is, the effective technique differs depending on the type of input data (which is also influenced by experimental conditions). Furthermore, even when using a single technique, bootstrapping is not always effective compared to the case where bootstrapping is not used. For example, in the case of MCC's N scheme, bootstrapping results in poor performance.

In the case of applying the ensemble technique, it can be seen that the accuracy is higher than that of the single technique. In general, when all three techniques are integrated (C & S & N), performance is improved compared to a single technique, and bootstrapping performance is improved compared to the case where no bootstrapping is used. Although it is shown that a single technique is used for a specific sample data or a combination of two techniques is better, there are three techniques that can not be known beforehand depending on the type of input data. The integrated analysis technique is generally regarded as significant.

FIG. 5 (B) shows an example of a comparison between the effects of the proposed technique and the conventional analysis technique. Conventional ensemble techniques were performed using Pearson, IDA and Lasso (see Le TD, Zhang J, Liu L, Li J. Ensemble methods for miRNA target prediction from expression data. PLoS One. The proposed method is compared with the results of ensemble of Pearson (P), IDA (I) and Lasso (L). In the proposed scheme, the performance of single direct correlation estimation (Single) is lower than that of the conventional scheme. However, in the case of bootstrapping with a single technique (Bootstrap), in the case of integrating the results of three techniques (Ensemble), when combining the results of three techniques together with bootstrapping (Bootstrap & Ensemble) This is excellent. Especially, the effect of applying bootstrapping and ensemble is the highest.

The present embodiment and drawings attached hereto are only a part of the technical idea included in the above-described technology, and it is easy for a person skilled in the art to easily understand the technical idea included in the description of the above- It will be appreciated that variations that may be deduced and specific embodiments are included within the scope of the foregoing description.

200: miRNA-mRNA network construction device
210: gene expression DB
220: Computer device
300: miRNA-mRNA network construction device
310: input device
320:
330: Storage device
340: Output device

Claims (12)

The computer device generating sample data including expression data of a plurality of miRNAs and expression data of a plurality of mRNAs;
The computer device generating a plurality of first association models for at least one of the plurality of miRNAs and at least one of the plurality of mRNAs; And
And the computer device integrating the plurality of first association models to generate a second association model. The method according to claim 1,
Wherein the computer device identifies information differentiated from the expression data of the miRNA and the expression data of the mRNA, and performs miRNA-mRNA association analysis to generate the sample data by integrating expression data of the identified miRNA and expression data of the mRNA Way. The method according to claim 1,
Wherein the computer device generates the first association model using at least one of a partial correlation coefficient estimation method, SPACE (Sparse Parental Correlation Estimation) and network deconvolution. The method according to claim 1,
Wherein the computer apparatus applies the partial correlation coefficient estimation method, SPACE (Sparse Parental Correlation Estimation) and network deconvolution to the sample data to generate the plurality of first association degree models. The method according to claim 1,
Wherein the computer device performs bootstraping of the sample data to generate a plurality of sets of sample data,
A plurality of correlation models are generated by applying at least one of a partial correlation coefficient estimation method, SPACE (Sparse Parental Correlation Estimation), and network deconvolution, which is an association degree estimation technique, to the plurality of sample data sets, Further comprising merging a plurality of association models generated by the estimation technique into a single model using a reverse rank based ensemble technique. The method according to claim 1,
Wherein the computer device integrates the plurality of first association models into the second association model using a reverse order based ensemble technique. The method according to claim 1,
Wherein the computer device further comprises generating a miRNA-mRNA network using the second association model. A computer-readable recording medium recording a program for causing a computer to execute the miRNA-mRNA association assay method according to any one of claims 1 to 7. An input device for receiving expression data of a plurality of miRNAs and expression data of a plurality of mRNAs;
a storage device for storing a program for building a miRNA-mRNA network; And
Executing the program using expression data of the plurality of miRNAs and expression data of a plurality of mRNAs as input data to generate expression data of the plurality of miRNAs and expression data of a plurality of mRNAs and sample data including input data , Generating a plurality of first association models for at least one of the plurality of miRNAs and at least one of the plurality of mRNAs using the sample data, integrating the plurality of first association models, MRNA network comprising at least one miRNA-mRNA network that generates a miRNA-mRNA network representing at least one of the plurality of miRNAs and at least one of the plurality of mRNAs based on the second association model, Network generating device. 10. The method of claim 9,
The arithmetic unit identifies information differentiated from the miRNA expression data and the expression data of the mRNA and integrates expression data of the identified miRNA and expression data of the mRNA to generate the sample data. . 10. The method of claim 9,
Wherein the computing device applies partial correlation coefficient estimation, SPACE, and network deconvolution to the sample data to generate the plurality of first association models. 10. The method of claim 9,
Wherein the computing device is further configured to integrate the plurality of first association models into the second association model using a reverse rank based ensemble technique, wherein the plurality of first association models are associated with the same edge in the plurality of first association models. To a final weight of the same edge, and integrates into the second association model.

Images