KR101289948B1 - Method and apparatus for classification of ligand by using molecular vibrational frequency patterns - Google Patents

Abstract

The present invention includes the steps of obtaining three-dimensional structural data information of the candidate ligand; Geometrically optimizing the data to convert the structure of the candidate ligand into a structure with a minimum energy level; Calculating the frequencies of candidate ligand molecules of the converted minimum energy level state structure; And predicting the action of a candidate ligand using the frequency.
According to the ligand classification method of the present invention, the molecular frequency pattern inherent to the ligand molecule can be used to easily predict and analyze whether the ligand acts as an agonist or an antagonist when the ligand acts on the actual receptor. The behavior of various candidate ligands can be predicted without going through experiments.

Description

Ligand classification method and apparatus using molecular frequency pattern {Method and Apparatus for Classification of Ligand by Using Molecular Vibrational Frequency Patterns}

The present invention relates to a ligand classifying method using a molecular frequency pattern, a classifying apparatus, and a computer readable medium recording a ligand classifying program.

G protein-coupled receptors (GPCRs) are a large group of membrane receptors that activate G-proteins in response to stimulation of the operator, resulting in the cyclic AMP / protein kinase A cascade. Activate at least one signaling cascade. This large population of receptors is widely found in organisms from yeast to human beings and is involved in, for example, the transformation of hormones, nerves and olfactory signals.

G protein-coupled receptors (GPCRs) have been shown to act as more than 25% of the receptors on the market for human drugs. Pharmaceuticals targeting GPCRs are a major product family with a market share of 40% of the overall pharmaceutical industry market.

GPCR is a transmembrane protein, consisting of seven membrane segments connected by three intercellular and three extracellular loops of varying lengths. It has been found that the crystal structure of binding ligands of GPCRs is associated with pockets and extracellular loops that bind to ligands (Cherezov V, Rosenbaum DM, Hanson MA, Rasmussen SG, Thian FS, Kobilka TS, et al. structure of an engineered human beta2-adrenergic G protein-coupled receptor.Science. 2007; 318 (5854): 1258-65 .; Shimamura T, Shiroishi M, Weyand S, Tsujimoto H, Winter G, Katritch V, et al. of the human histamine H1 receptor complex with doxepin.Nature. 2011; 475 (7354): 65-70). Ligands in GPCRs activate receptors through a mechanism that changes the receptor structure to the active form.

The molecular mechanism of ligand-receptor interactions of GPCRs is not yet known and many studies are currently being conducted. To explain the ligand-receptor interaction mechanism, various theories such as shape theory, binding theory and vibration theory have been proposed (Keller A, Vosshall LB. A psychophysical test of the vibration theory of olfaction. Nat Neurosci. 2004; 7 (4): 337-8 .; Takane SY, Mitchell JB.A structure-odour relationship study using EVA descriptors and hierarchical clustering.Org Biomol Chem. 2004; 2 (22): 3250-5; Korean Patent Publication 10-2005 -0029109).

However, ligands that act on GPCRs, such as histamine receptors or adenosine receptors, are highly diverse in structure, and there are limitations in predicting the effect of ligands by structural analysis. In addition, when actual in vitro experiments are performed, it is troublesome to go through a multi-step preparation process such as separation, filtration and washing. Therefore, there is a need for a faster and simpler method for predicting the effect of GPCR ligands.

Accordingly, the inventors of the present invention have studied to predict and classify the effect of G-protein-linked receptor ligands more quickly and simply. As a result, it is possible to easily predict whether the ligands are only active by analyzing molecular frequency patterns inherent to the ligand molecules. It confirmed and completed this invention.

Accordingly, the present invention provides a computer-readable medium recording a ligand classification method, a classification device, and a ligand classification program using a molecular frequency pattern.

In order to solve the above problems, the present invention

Obtaining three-dimensional structural data information of the candidate ligand; Geometrically optimizing the data to convert the structure of the candidate ligand into a structure with a minimum energy level; Calculating the frequencies of candidate ligand molecules of the converted minimum energy level state structure; And it provides a ligand classification method comprising the step of predicting the action of the candidate ligand using the frequency.

In another aspect, the present invention provides an input unit for receiving the three-dimensional structure data information of the candidate ligand; A first calculation unit which geometrically optimizes the data inputted to the input unit to convert the structure of the candidate ligand into a structure of a minimum energy level state; A second operation unit for calculating a frequency of candidate ligand molecules of the converted minimum energy level state structure; A third operation unit predicting the action of the candidate ligand using the frequency; And an output unit for outputting the action of the candidate ligand predicted through the third operation unit.

In addition, the present invention comprises the steps of receiving the three-dimensional structure data information of the candidate ligand through a computer input device; Causing the computer computing device to geometrically optimize the data to convert the structure of the candidate ligand into a structure of a minimum energy level; Causing a computer computing device to calculate frequencies of candidate ligand molecules of the transformed minimum energy level state structure; Causing a computing device of the computer to predict the action of a candidate ligand using the frequency; And a computer program for recording a ligand classification program for executing a step of causing an output device of the computer to output the action of the candidate ligand predicted by the operation residual.

According to the ligand classification method and classification apparatus of the present invention, by using the molecular frequency pattern inherent to the ligand molecule, it is possible to easily predict and analyze whether the ligand acts as an agonist or an antagonist when the ligand acts on the actual receptor, and thus various candidate ligands. Whether or not can be predicted without going through experiments.

1 is a flowchart illustrating a ligand classification method according to the present invention.
Figure 2 is a block diagram showing the configuration of a ligand classification apparatus according to the present invention.
3 is an example of molecular frequency obtained in accordance with one embodiment of the present invention.
Figure 4 shows a schematic of the histamine receptor binding ligand according to an embodiment of the present invention.
Figure 5 shows a schematic of the adenosine receptor binding ligand according to an embodiment of the present invention.

The present invention includes the steps of obtaining three-dimensional structural data information of the candidate ligand; Geometrically optimizing the data to convert the structure of the candidate ligand into a structure with a minimum energy level; Calculating the frequencies of candidate ligand molecules of the converted minimum energy level state structure; And it provides a ligand classification method comprising the step of predicting the action of the candidate ligand using the frequency.

In one embodiment of the present invention, predicting the action of the candidate ligand may be a step of obtaining a schematic by comparing the similarity of the frequencies of the candidate ligand and the present ligand molecule.

In another embodiment of the present invention, predicting the action of the candidate ligands comprises: dividing the molecular frequencies of the candidate ligands at regular intervals; Summing the intensity of molecular frequencies present in the same interval section to form a representative value of the section; And comparing the similarity with the ligand based on the pattern of the representative value to obtain a systematic diagram.

In another embodiment of the present invention, the ligand may be a rdopsin-like G-protein-coupled receptor (GPCR) binding ligand, more specifically the rhodopsin-like G-protein G protein-coupled receptors (GPCRs) may be, but are not limited to, histamine receptors or adenosine receptors.

In the present invention, the 'candidate ligand' is a ligand to be analyzed to know the effect of the action, and the 'present ligand' means a ligand known to bind to the original receptor. In one embodiment of the present invention, when the receptor is a histamine receptor, the histamine is the present ligand, and when the receptor is the adenosine receptor, the ligand is adenosine. The present invention compares the frequency of the candidate ligand with the present ligand and predicts the effect of the candidate ligand through its similarity. If the frequency pattern with the present ligand appears similar and is located adjacent to the tree, the candidate ligand can be expected to act as an agonist and vice versa.

1 is a flow chart illustrating a ligand classification method according to an embodiment of the present invention.

Each step will be described in detail below.

Acquiring the three-dimensional structure data information of the candidate ligand may be performed by downloading from a website providing general information about the candidate ligand, and more specifically, it may be downloaded from NCBI's PubChem Compound Database. It is not limited.

Since the three-dimensional structure data information obtained by the above method does not provide a structure of the minimum energy state, then a geometric optimization step of the candidate ligands for converting the structure of the candidate ligands into the minimum energy state is performed. This can be done via Restricted Hartree-Fock calculations using, but not limited to, the GAMESS program package.

The frequency of the candidate ligand molecules of the transformed minimum energy level state structure is then measured. Thereafter, in order to more easily compare the frequencies, dividing the molecular frequencies of candidate ligands at regular intervals; A step of adding the intensities of molecular frequencies present in the same interval section to represent the representative value of the section (corralled intensity of molecular vibrational frequency, CIMVF) is performed.

The interval (corral interval) may be set to an arbitrary value for convenience of calculation, more specifically, to a natural number within the range of 3 to 50, preferably to 5 to 10 for the accuracy of the result. It may be set but is not limited thereto.

Thereafter, a step of obtaining a systematic diagram by comparing the similarity with the ligand based on the pattern of the representative value is performed.

In order to compare the similarity of ligands, an application such as agglomerative hierarchical clustering (complete linkage) application such as J-Express can be used, and a tree diagram can be obtained through a conventional method using a program such as TreeView.

When the candidate is located close to the present ligand in the schematic, it can be predicted that the candidate ligand acts as an agonist and vice versa.

The present invention also includes an input unit for receiving three-dimensional structural data information of the candidate ligand; A first calculation unit which geometrically optimizes the data inputted to the input unit to convert the structure of the candidate ligand into a structure of a minimum energy level state; A second operation unit for calculating a frequency of candidate ligand molecules of the converted minimum energy level state structure; A third operation unit predicting the action of the candidate ligand using the frequency; And an output unit for outputting the action of the candidate ligand predicted through the third operation unit.

Figure 2 is a block diagram showing the configuration of a ligand classification apparatus according to the present invention.

The ligand classification apparatus according to the present invention includes an input unit 10, a calculation unit 20, and an output unit 30. In addition, the operation unit 20 includes a first operation unit 21, a second operation unit 22, and a third operation unit 23.

The input unit 10 may use an input device such as a keyboard or a mouse. In addition, the input unit 10 may include a device for measuring the three-dimensional structure of the candidate ligand, from which the three-dimensional structure data of the candidate ligand can be input.

The calculation unit 20 calculates and predicts the possibility of acting as an agonist / antagonist of the candidate ligand by using the data input from the input unit 10. The calculating unit 20 may be implemented by a microcomputer or a computer.

In one embodiment of the present invention, the calculation unit 20 includes a first calculation unit 21 for geometrically optimizing the data input to the input unit to convert the structure of the candidate ligand into a structure of the minimum energy level state; A second calculation unit 22 for calculating a frequency of candidate ligand molecules of the converted minimum energy level state structure; And a third operation unit 23 for predicting the action of the candidate ligand using the frequency.

In another embodiment of the present invention, the third operation unit 23 divides the molecular frequency of the candidate ligand at regular intervals; Summing the intensity of molecular frequencies present in the same interval section to form a representative value of the section; And comparing the similarity with the present ligand based on the pattern of the representative value to obtain a phylogenetic tree.

In another embodiment of the present invention, the ligand may be a rdopsin-like G-protein-coupled receptor (GPCR) binding ligand, more specifically the rhodopsin-like G-protein G protein-coupled receptors (GPCRs) may be, but are not limited to, histamine receptors or adenosine receptors.

The output unit 30 outputs the action of the candidate ligand predicted by the calculating unit 20. The output unit 30 is not particularly limited thereto, but may be implemented as an output device such as a monitor or an LCD panel.

The present invention can also be embodied as a recording medium produced by a computer program and stored therein so that the ligand classification method can be performed on a computer.

That is, the computer-readable medium in which the ligand classification program according to the present invention is recorded,

Receiving 3D structural data information of the candidate ligand through a computer input device;

Causing the computer computing device to geometrically optimize the data to convert the structure of the candidate ligand into a structure of a minimum energy level;

Causing a computer computing device to calculate frequencies of candidate ligand molecules of the transformed minimum energy level state structure;

Causing a computing device of the computer to predict the action of a candidate ligand using the frequency; And

A ligand classification program can be recorded for executing the step of causing the output device of the computer to output the action of the candidate ligand predicted by the computation residual.

BEST MODE FOR CARRYING OUT THE INVENTION Hereinafter, the present invention will be described in detail with reference to the following examples. However, the following examples are intended to illustrate the contents of the present invention, but the scope of the present invention is not limited to the following examples. Embodiments of the present invention are provided to more fully describe the present invention to those skilled in the art.

< Example  1> histamine receptor binding Ligand  Classification

1. Data set

The SMILES and structure data format (SDF) files of the data set were downloaded from the PubChem Compound Database of the National Center for Biotechnology Information (NCBI) and used for later analysis. All 47 ligand molecules in the data set included 9 histamine agonists and 38 histamine antagonists, which are shown in Table 1 below.

PubChem ID Compound name Receptor Function type PubChem ID Compound name Receptor Function type 87653 2-thiazoleethanamine HRH1 agonist 3957 loratadine HRH1 antagonist 2366 betahistine HRH1 agonist 4034 meclozine HRH1 antagonist 75919 demethylbetahistine HRH1 agonist 4761 pheniramine HRH1 antagonist 7741 betazole HRH2 agonist 4927 promethazine HRH1 antagonist 3077 dimaprit HRH2 agonist 5002 quetiapine HRH1 antagonist 3692 imetit HRH3 agonist 3032915 burimamide HRH2 antagonist 126688 amthamine HRH3, H4 agonist 2756 cimetidine HRH2 antagonist 41376 impromidine HRH4 agonist 5282136 lafutidine HRH2 antagonist 5227 SKF91488 HR agonist 3033637 nizatidine HRH2 antagonist 2267 azelastine HRH1 antagonist 5282450 pibutidine HRH2 antagonist 2678 cetirizine HRH1 antagonist 3001055 ranitidine HRH2 antagonist 2725 chlorpheniramine HRH1 antagonist 5105 roxatidine HRH2 antagonist 26987 clemastine HRH1 antagonist 50287 tiotidine HRH2 antagonist 6726 cyclizine HRH1 antagonist 9954017 A-349821 HRH3 antagonist 124087 desloratadine HRH1 antagonist 9818903 ABT-239 HRH3 antagonist 33036 dexchlorpheniramine HRH1 antagonist 2366 betahistine HRH3 antagonist 21855 dimetindene HRH1 antagonist 6422124 ciproxifan HRH3 antagonist 3100 diphenhydramine HRH1 antagonist 3035746 iodophenpropit HRH3 antagonist 667477 doxepin HRH1 antagonist 9948102 pitolisant HRH3 antagonist 3162 doxylamine HRH1 antagonist 3035905 thioperamide HRH3, H4 antagonist 3191 ebastine HRH1 antagonist 2790 clobenpropit HRH3 antagonist 19105 embramine HRH1 antagonist 4908365 JNJ-7777120 HRH4 antagonist 3348 fexofenadine HRH1 antagonist 10446295 VUF6002 HRH4 antagonist 1549000 levocetirizine HRH1 antagonist

2. Geometric Optimization ( Geometry optimization ) And molecular frequency calculation

In order to calculate the molecular frequency, the structure of the compound must be geometrically optimized. Since the theoretical three-dimensional shape SDF is not the minimum energy and does not represent the lowest energy form in the vacuum, solvent, or bond pocket, each SDF file of the ligand molecules is in a single low-energy form using the GAMESS program package. Has gone through the transition process. Restricted Hartree-Fock calculations were performed to geometrically optimize the molecules using the BLYP DFT method with the 6-31G base set. Although the calculation of molecular frequencies is related to molecular morphology, each result represents a molecular morphology. Each geometric optimization result was used to calculate molecular frequencies with the HESSIAN RUNTYP of the GAMESS program. An example of the obtained molecular frequency is shown in FIG.

3. Cluster strength of molecular frequency ( Corralled Intensity of  Molecular Vibrational Frequency , CIMVF )of Genealogical  Clustering Hierarchical clustering )

To simplify molecular comparisons, the calculated molecular frequencies were sorted in ascending order and divided by 5 intervals. By summing the each intensity in each frequency interval from 0 to 5000 cm ^-, at a frequency range of ¹ it was the representative value for each interval. For the final result, the potential molecular indicators for each molecule were represented as one-dimensional vectors containing 1000 elements. Finally, a similarity matrix comprising indicators of the 47 ligands of the histamine receptor was hierarchically clustered. In this example, the similarity matrix finally clustered into trees with 47 vertices. CIMVF calculations were performed by in-house scripts written by Python. 4 is a schematic diagram through hierarchical clustering of the similarity matrix through CIMVF.

As shown in FIG. 4, eight agents except for impromidine were located near histamine (indicated by dashed circles) and all antagonists were clustered close to each other in the radial tree near histamine. Thus, it was found that there is a regional difference between agonists and inhibitors in the tree, and that molecular frequencies can provide information on the agonist / inhibitor classification for histamine receptors.

< Comparative example  1> histamine receptor Ligand Structure comparison Through Ligand  Classification

Similarity of the molecular structure between histamine and other 46 compounds was analyzed from the SMILES of the compounds. Structural similarity was calculated, expressed as the Tanimoto distance of each molecule from histamine. Tanimoto coefficients for pairwise comparisons of molecules are the most widely used measure of molecular structure similarity, and Tc = N _ab / ( N _a + N _b N _ab ). Where N _ab Is the number of common bits using molecular fingerprint, N _a Is the unique bit of the molecule a, N _b Is the unique bit of the molecule b. In this example, molecular similarity was calculated with Tanimoto coefficients using a 38-bit set.

Tanimoto coefficients of histamine and other ligand molecules are shown in Table 2 below.

Compound Tanimoto coefficient Function type Compound Tanimoto coefficient Function type histamine One agonist imetit 0.45 agonist loratadine 0.13 antagonist burimamide 0.43 antagonist JNJ-7777120 0.13 antagonist thioperamide 0.41 antagonist pitolisant 0.12 antagonist iodophenpropit 0.3 antagonist embramine 0.12 antagonist demethylbetahistine 0.3 agonist cyclizine 0.11 antagonist ciproxifan 0.3 antagonist dimetindene 0.11 antagonist impromidine 0.29 agonist meclozine 0.11 antagonist clobenpropit 0.29 antagonist nizatidine 0.1 antagonist betahistine 0.27 antagonist promethazine 0.1 antagonist amthamine 0.25 agonist azelastine 0.1 antagonist 2-thiazoleethanamine 0.25 agonist cetirizine 0.1 antagonist cimetidine 0.24 antagonist levocetirizine 0.1 antagonist betazole 0.22 agonist tiotidine 0.09 antagonist VUF6002 0.19 antagonist ABT-239 0.09 antagonist doxylamine 0.18 antagonist doxepin 0.09 antagonist pheniramine 0.18 antagonist dimaprit 0.08 agonist chlorpheniramine 0.18 antagonist A-349821 0.08 antagonist dexchlorpheniramine 0.18 antagonist pibutidine 0.08 antagonist fexofenadine 0.16 antagonist SKF91488 0.08 agonist desloratadine 0.14 antagonist roxatidine 0.08 antagonist diphenhydramine 0.14 antagonist ranitidine 0.07 antagonist clemastine 0.13 antagonist lafutidine 0.07 antagonist ebastine 0.13 antagonist quetiapine 0.07 antagonist

In general, when the Tanimoto coefficient is> 0.85, it means that there is a very high similarity, and when> 0.75, it means that the material has similar biologically active properties with a similar purpose as the cluster molecule. The highest Tanimoto modulus among the agonists was 0.45 (imetit), with no similarity as potential agents. In addition, the lowest Tanimoto coefficient value among the agonists was 0.08 (dimaprit and SKF91488), and was also low as an antagonist.

Therefore, the relationship between the Tanimoto coefficient and the function of the molecule was found to be insignificant, and it was found that it is difficult to predict the biological activity of the ligand through the structure of the ligand.

< Example  2> adenosine receptor binding Ligand  Classification

SMILES and three-dimensional structure data format (SDF) files of adenosine receptor ligand data sets were downloaded from the PubChem Compound Database of the National Center for Biotechnology Information (NCBI) and used for later analysis. The data sets are shown in Table 3 below.

PubChem ID Compound name Receptor Function type PubChem ID Compound name Receptor Function type 60961 adenosine AdoR agonist 1329 DPCPX AdoRA1 antagonist 10475082 2-MeCCPA AdoRA1 agonist 6439091 FK453 AdoRA1 antagonist 16218845 ADAC AdoRA1 agonist 9821511 FR194921 AdoRA1 antagonist 123807 CCPA AdoRA1 agonist 64627 KW-3902 AdoRA1 antagonist 4402 CPA AdoRA1 agonist 9953065 SLV320 AdoRA1 antagonist 158795 CVT-510 AdoRA1 agonist 9902054 WRC-0571 AdoRA1 antagonist 11957545 ENBA AdoRA1 agonist 2690 CGS15943 AdoRA2A antagonist 130214 GR79236 AdoRA1 agonist 6603761 CSC AdoRA2A antagonist 104795 NECA AdoRA1 agonist 5311037 KW-6002 AdoRA2A antagonist 9936084 NNC21-0136 AdoRA1 agonist 11223144 MSX-2 AdoRA2A antagonist 93205 R-PIA (L-PIA) AdoRA1 agonist 10668061 SCH442416 AdoRA2A antagonist 5311431 S-EBNA AdoRA1 agonist 176408 SCH58261 AdoRA2A antagonist 164305 SDZWAG994 AdoRA1 agonist 2153 theophylliine AdoRA2A antagonist 3086599 CGS21680 AdoRA2A agonist 9949541 VER-6947 AdoRA2A antagonist 9841284 CHA AdoRA2A agonist 9923269 VER-7835 AdoRA2A antagonist 1585 CV1808 AdoRA2A agonist 5697 XAC AdoRA2A antagonist 219024 CVT-3146 AdoRA2A agonist 176407 ZM241385 AdoRA2A antagonist 164437 HENECA AdoRA2A agonist 5372720 alloxazine AdoRA2B antagonist 131432 SHA40 AdoRA2A agonist 1882 compd19a AdoRA2B antagonist 9833519 UK-432097 AdoRA2A agonist 10196114 CVT-5440 AdoRA2B antagonist 9576912 WRC-0470 (bino) AdoRA2A agonist 1676 enprofylline AdoRA2B antagonist 8974 2-CADO AdoRA2B agonist 11716665 LAS-38096 AdoRA2B antagonist 11717831 BAY60-6583 AdoRA2B agonist CSID_19920712 MRE2029-F20 AdoRA2B antagonist 9975036 LUF5835 AdoRA2B agonist 6603931 MRS1754 AdoRA2B antagonist 2519 caffeine AdoR antagonist 5311041 OSIP-339391 AdoRA2B antagonist 158540 BG9719 AdoRA1 antagonist 11689583 PSB-36 AdoRA2B antagonist 216466 BG9928 AdoRA1 antagonist

Thereafter, steps 2 and 3 of Example 1 were performed in the same manner, and the resulting system diagram is shown in FIG. 5. According to Figure 5, except for ZM241385 it can be seen that the antagonists (tagonists) denoted by t all form a different group from adenosine, the method of the present invention can distinguish between antagonists / agents with a high probability. Could know.

Having described the specific parts of the present invention in detail, it will be apparent to those skilled in the art that such specific descriptions are merely preferred embodiments, and thus the scope of the present invention is not limited thereto. will be. Accordingly, the actual scope of the present invention will be defined by the appended claims and their equivalents.

Claims (10)

Obtaining three-dimensional structural data information of the candidate ligand;
Geometrically optimizing the data to convert the structure of the candidate ligand into a structure with a minimum energy level;
Calculating the frequencies of candidate ligand molecules of the converted minimum energy level state structure; And
Predicting the action of a candidate ligand using the frequency. The method of claim 1,
Predicting the action of the candidate ligand, Ligand classification method to obtain a phylogenetic tree by comparing the similarity of the frequency of the candidate ligand and the present ligand molecule. The method of claim 1,
Predicting the action of the candidate ligand,
Dividing the molecular frequency of the candidate ligand by regular intervals; Summing the intensity of molecular frequencies present in the same interval section to form a representative value of the section; And comparing the similarity with the present ligand based on the pattern of the representative value to obtain a systematic diagram. 4. The method according to any one of claims 1 to 3,
Wherein said ligand is a rhodopsin-like G protein-coupled receptor (GPCR) binding ligand. 5. The method of claim 4,
And said rhodopsin-like G protein-coupled receptor (GPCR) is a histamine receptor or an adenosine receptor. An input unit configured to receive 3D structural data information of a candidate ligand;
A first calculation unit which geometrically optimizes the data inputted to the input unit to convert the structure of the candidate ligand into a structure of a minimum energy level state;
A second operation unit for calculating a frequency of candidate ligand molecules of the converted minimum energy level state structure;
A third operation unit predicting the action of the candidate ligand using the frequency; And
Ligand classification apparatus comprising an output unit for outputting the action of the candidate ligand predicted through the third operation unit. The method according to claim 6,
The third operation unit divides the molecular frequency of the candidate ligand at regular intervals; Summing the intensity of molecular frequencies present in the same interval section to form a representative value of the section; And comparing the similarity with the present ligand based on the pattern of the representative value to obtain a phylogenetic tree. 8. The method according to claim 6 or 7,
Wherein said ligand is a rdopsin-like G protein-coupled receptor (GPCR) binding ligand. 9. The method of claim 8,
Wherein said rhodopsin-like G protein-coupled receptor (GPCR) is a histamine receptor or an adenosine receptor. Receiving 3D structural data information of the candidate ligand through a computer input device;
Causing the computer computing device to geometrically optimize the data to convert the structure of the candidate ligand into a structure of a minimum energy level;
Causing a computer computing device to calculate frequencies of candidate ligand molecules of the transformed minimum energy level state structure;
Causing a computing device of the computer to predict the action of a candidate ligand using the frequency; And
And a ligand classification program for executing a step of causing a computer output device to output the action of a candidate ligand predicted by the computation residual.

Images