Computer program, information processing apparatus, and information processing method

Abstract

To provide a computer program, an information processor and an information processing method for screening a drug candidate compound at a low calculation cost.SOLUTION: The computer program causes a computer to execute processing of acquiring binding information between a protein molecule and a compound molecule obtained by molecular docking simulation, and generating an evaluation index for screening whether or not the compound molecule is a candidate compound interacting with the protein molecule as a target protein based on the acquired binding information.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a computer program, an information processing device, and an information processing method. [Background technology]

[0002] Evaluation of intermolecular interactions between biopolymers is fundamental for understanding biological phenomena and advancing drug development. Most interaction evaluations rely on experimental methods or high-precision computational methods, which are time-consuming and costly. The number of intermolecular interactions between biomolecules and drugs is enormous, and establishing a method for evaluating these interactions efficiently and at low cost is an important challenge.

[0003] One example of interaction simulation using computational science is molecular docking simulation. Molecular docking simulation inputs the three-dimensional structures of a protein and a compound, and predicts the predicted complex structure between the compound and the protein and the activity score corresponding to that predicted complex structure. On the other hand, there is molecular dynamics simulation, which can evaluate how the three-dimensional structure of a molecule changes over time.

[0004] Patent Document 1 discloses a method for constructing the three-dimensional structure of a protein-ligand complex by applying molecular dynamics simulation using the three-dimensional structure of a ligand docked to a receptor protein model. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Patent No. 4315960 Summary of the Invention [Problem to be solved by the invention]

[0006] Generally, to accurately evaluate interactions, molecular dynamics simulations must be performed for times on the order of μsec to msec, and using these simulations to screen drug candidate compounds for thousands to tens of thousands of compound-protein pairs is impractical from the perspective of computational cost.

[0007] The present invention has been made in view of the above circumstances, and has as its object to provide a computer program, an information processing device, and an information processing method that are capable of screening drug candidate compounds at low calculation costs. [Means for solving the problem]

[0008] The present application includes a plurality of means for solving the above-mentioned problems. As one example, a computer program causes a computer to execute a process of acquiring binding information between a protein molecule and a compound molecule obtained by a molecular docking simulation, and generating, based on the acquired binding information, an evaluation index for screening whether or not the compound molecule is a candidate compound that interacts with the protein molecule as a target protein. [Effects of the Invention]

[0009] According to the present invention, drug candidate compounds can be screened at low computational cost. [Brief explanation of the drawings]

[0010] [Figure 1] FIG. 1 is a diagram illustrating an example of a configuration of an information processing apparatus according to an embodiment of the present invention. [Figure 2] FIG. 10 is a diagram illustrating an example of processing by a molecular docking simulator. [Figure 3] FIG. 10 is a diagram illustrating an example of processing by a molecular dynamics simulator. [Figure 4] FIG. 10 illustrates an example of processing performed by an information processing device. [Figure 5] FIG. 1 is a diagram showing an example of a change in RMSD over time. [Figure 6]FIG. 10 is a diagram illustrating an example of a dataset for binding pose validation. [Figure 7] An example of a binding pose output from a molecular docking simulation is shown. [Figure 8] FIG. 1 shows an example of a binding pose labeled according to whether the binding pose is correct or incorrect. [Figure 9] FIG. 10 is a diagram showing an example of an evaluation result by molecular dynamics simulation within a predetermined time period. [Figure 10] FIG. 10 is a diagram showing an example of a confusion matrix for determining whether a binding pose is correct or incorrect. [Figure 11] FIG. 10 is a diagram showing an example of ROC-AUC for determining whether a binding pose is correct or incorrect. [Figure 12] FIG. 1 shows an example of target compounds labeled according to activity value. [Figure 13] FIG. 10 is a diagram showing an example of a confusion matrix for determining the presence or absence of activity. [Figure 14] FIG. 10 is a diagram showing an example of ROC-AUC for determining the presence or absence of activity. [Figure 15] FIG. 10 is a diagram showing an example of statistical values ​​of RMSD. [Figure 16] FIG. 10 is a diagram illustrating an example of a processing procedure of an information processing device. DETAILED DESCRIPTION OF THE INVENTION

[0011] The present invention will be described below with reference to the drawings illustrating embodiments thereof. FIG. 1 is a diagram showing an example of the configuration of an information processing device 50 according to this embodiment. The information processing device 50 can be connected to a molecular docking simulator 10, a molecular dynamics simulator 20, and a database 30 via a communication network 1. The database 30 stores protein data 31 and compound data 32. The information processing device 50 can execute a molecular docking simulation by accessing the molecular docking simulator 10. The information processing device 50 can also execute a molecular dynamics (MD) simulation by accessing the molecular dynamics simulator 20. The information processing device 50 can acquire the protein data 31 and compound data 32 by accessing the database 30. Note that the information processing device 50 may be configured with a plurality of information processing devices, for example, with functions distributed among them.

[0012] The information processing device 50 includes a control unit 51 that controls the entire device, a communication unit 52, a memory 53, a display unit 54, an operation unit 55, a storage unit 56, and a recording medium reading unit 58.

[0013] The control unit 51 may be configured by incorporating a required number of central processing units (CPUs), micro-processing units (MPUs), graphics processing units (GPUs), etc. The control unit 51 may also be configured by combining digital signal processors (DSPs), field-programmable gate arrays (FPGAs), etc.

[0014] The communication unit 52 includes a communication module and has a function of communicating with the molecular docking simulator 10, the molecular dynamics simulator 20, and the database 30 via the communication network 1.

[0015] The display unit 54 is configured with a liquid crystal display, an organic EL display, or the like, and provides a UI (user interface) to the user by displaying required information. The display unit 54 is equipped with a microphone and a speaker, and can input and output audio. Note that an external display device may be used instead of the display unit 54.

[0016] The operation unit 55 is configured, for example, by a touch panel, and can operate icons displayed on the display unit 54, move and operate a cursor, input characters, etc. The operation unit 55 may be configured by buttons, switches, etc., or may be configured by a keyboard, mouse, etc. The operation unit 55 provides a UI to the user by accepting user operations. Note that an external terminal device for operation may be provided instead of the operation unit 55.

[0017] The storage unit 56 can be configured with a semiconductor memory or a hard disk, and stores a computer program (program product) 57 and required information.

[0018] The computer program 57 can be stored in the storage unit 56 by reading the computer program 57 recorded on a recording medium (for example, an optically readable disk storage medium such as a CD-ROM) M using the recording medium reading unit 58. The computer program 57 may also be stored in the storage unit 56 by reading the computer program 57 recorded on a recording medium such as a storage device (semiconductor memory such as a solid state drive (SSD)) connected according to a standard for connecting to a computer (for example, USB (Universal Serial Bus) or other standard). Alternatively, the computer program 57 may be downloaded from an external device via the communication unit 52 and stored in the storage unit 56.

[0019] The memory 53 can be configured with a semiconductor memory such as a static random access memory (SRAM), a dynamic random access memory (DRAM), or a flash memory. A computer program 57 can be loaded into the memory 53, and the control unit 51 can execute the computer program 57. The control unit 51 can execute processing defined by the computer program 57. In other words, processing by the control unit 51 is also processing by the computer program 57.

[0020] FIG. 2 is a diagram illustrating an example of processing by the molecular docking simulator 10. Molecular docking simulation using the molecular docking simulator 10 is a computational method for predicting the interaction between two biomolecules that form a complex, and predicts the binding energy and binding structure of a protein-ligand (compound) complex structure. Predicting the protein-ligand complex structure is called "docking" and plays an important role in drug discovery based on the three-dimensional structure of biomolecules. As shown in FIG. 2, protein data and compound data are input to the molecular docking simulator 10. The protein data and compound data are, for example, the molecular coordinates of the protein and compound, respectively. Using the input molecular coordinates of the protein and compound, the molecular docking simulator 10 performs a conformational search of the compound (compound molecule) within the protein binding pocket, evaluates the molecular affinity (binding energy) between the compound molecule and the protein using a score function, and outputs binding information for the compound (e.g., a binding pose and an activity score at the binding pose). Note that, in this specification, a "binding pose" (also referred to as a "docking pose") refers to a predicted complex structure between a protein molecule and a compound molecule. Hereinafter, the predicted complex structure will be described as a "binding pose." Multiple pieces of binding information may be output. That is, the molecular docking simulator 10 can output multiple binding poses and their respective activity scores. The binding pose, also called a docking pose, is expressed by the position and orientation of the entire compound molecule in the protein binding pocket, the rotation angle of the covalent bond, etc. The activity score is a value expressed as a score that defines the interaction activity between molecules and indicates a predicted value of the interaction activity. The formula that defines the interaction activity between molecules can be set appropriately depending on the purpose, etc.

[0021] When a molecular docking simulation using the molecular docking simulator 10 is performed for screening purposes, the atomic coordinates of the protein are generally fixed and the protein is considered to be a rigid body. Although there are molecular docking simulations in which the atomic coordinates of the protein are slightly moved while docking, this method still results in high computational costs when performed on a large scale, such as for screening. Furthermore, even if a molecular docking simulation in which the atomic coordinates of the protein are slightly moved while docking is performed, it is not possible to reproduce the true dynamic behavior in solution. As a result, dynamic structural changes that are important for protein function are ignored. As a result, the binding poses and activity scores predicted by the molecular docking simulation are less accurate, and incorrect binding poses and incorrect activity scores are output.

[0022] The information processing device 50 can screen whether or not a compound molecule is a candidate compound that interacts with a protein molecule as a target protein by eliminating erroneous binding poses and erroneous activity scores.

[0023] 3 is a diagram showing an example of processing by the molecular dynamics simulator 20. The molecular dynamics simulation by the molecular dynamics simulator 20 is based on Newton's equations of motion and takes into account various forces acting between atoms. By inputting the initial structure of the molecule, the simulator simulates how the three-dimensional structure of the molecule changes over time and outputs the molecular structure at each time. The flow of the molecular dynamics simulation is as follows: (1) arrange the atoms, (2) calculate the forces acting on the atoms, and (3) calculate the movement of the atoms, and steps (1) to (3) are repeated.

[0024] As shown in FIG. 3, when positional information of each atom of a protein and a compound is input to the molecular dynamics simulator 20, the molecular dynamics simulator 20 outputs the movement of each atom.

[0025] Molecular dynamics simulations take into account the flexibility of proteins and can perform accurate simulations, but they are computationally expensive. Generally, to accurately determine intermolecular interactions, simulation execution times on the order of μsec to msec are required, making it unrealistic from the perspective of computational cost to use them for screening purposes for thousands to tens of thousands of compound-protein pairs, and therefore molecular dynamics simulations cannot be used in practice.

[0026] The information processing device 50 can screen whether a compound molecule is a candidate compound that interacts with a protein molecule as a target protein by performing a short-time molecular dynamics simulation. Here, "short time" means a time shorter than the time required for performing a molecular dynamics simulation, which is on the order of μsec to msec, and can be, for example, a time on the order of nsec.

[0027] 4 is a diagram showing an example of processing by the information processing device 50. The control unit 51 acquires protein data and compound data from the database 30. In this case, the protein is a target protein, and the compound (ligand) is a compound that is the subject of a determination of whether or not it binds to the target protein. The protein data and compound data are the molecular coordinates of the protein and the molecular coordinates of the compound. The control unit 51 inputs the protein data and compound data into the molecular docking simulator 10, and executes a molecular docking simulation to acquire a binding pose (an estimated complex structure between a protein molecule and a compound molecule).

[0028] The control unit 51 inputs the acquired binding pose into the molecular dynamics simulator 20 and executes a molecular dynamics simulation within a predetermined time. The input data includes positional information of each atom of the protein and the compound. The "predetermined time" is synonymous with the above-mentioned "short time" and can be, for example, a time on the order of nanoseconds. A molecular dynamics simulation within a predetermined time can be executed with low computational cost. By executing a molecular dynamics simulation within a predetermined time, the control unit 51 generates an evaluation index for screening whether a compound molecule is a candidate compound that interacts with a protein molecule as a target protein.

[0029] As shown in Figure 4, the evaluation index can be, for example, RMSD (Root Mean Square Deviation), which is defined as the square root of the average of the squares of the deviations between corresponding atoms of a compound and a protein.

[0030] As described above, the control unit 51 can acquire binding information between protein molecules and compound molecules obtained by molecular docking simulation, and based on the acquired binding information, generate an evaluation index for screening whether or not the compound molecule is a candidate compound that interacts with the protein molecule as a target protein.

[0031] By generating an evaluation index, it is not necessary to perform molecular dynamics simulations for times on the order of μsec to msec, and drug candidate compounds can be screened at low calculation costs.

[0032] More specifically, the control unit 51 acquires position information of each atom of each protein molecule and compound molecule obtained by molecular docking simulation, performs molecular dynamics simulation within a predetermined time, calculates statistical values ​​of root mean square deviations of the position information of the compound, and generates the temporal change of the calculated statistical values ​​within the predetermined time as an evaluation index. The "predetermined time" means a time shorter than the simulation time required for a general molecular dynamics simulation, which is on the order of μsec to msec, and can be, for example, on the order of nsec. The statistical value can be, for example, the maximum value of RMSD. Note that the statistical value is not limited to the maximum value of RMSD, and can be, for example, the average value of RMSD, the nth largest RMSD, or the maximum value of average RMSD over a certain period of time.

[0033] This eliminates the need to perform molecular dynamics simulations for times on the order of μsec to msec, and makes it possible to screen drug candidate compounds with high accuracy and at low calculation costs compared to general molecular dynamics simulations.

[0034] Figure 5 shows an example of the change in RMSD over time. Figure 5 is a schematic illustration of the change in RMSD over time and may differ from actual experimental results. In Figure 5, the vertical axis represents RMSD, and the horizontal axis represents the execution time of the molecular dynamics simulation. As shown in Figure 5, proteins and compounds that have a correct binding pose and bind with high affinity have a small, almost constant RMSD value that does not change. On the other hand, proteins and compounds that have an incorrect binding pose or no activity have a large change in RMSD value over time.

[0035] This allows the RMSD to be used as an evaluation index for screening whether a compound molecule is a candidate compound that interacts with a protein molecule as a target protein.

[0036] Next, we will explain whether RMSD as an evaluation index can distinguish between correct and incorrect binding poses and whether it can distinguish between active and inactive binding poses. First, we will explain why RMSD can distinguish between correct and incorrect binding poses.

[0037] Figure 6 shows an example of a dataset for binding pose validation. The dataset uses compound-protein pairs whose complex structures are already known through experiments. Examples of proteins include ERK2, p38, EGFR, PTPN1, PPARG, and BACE1. A molecular docking simulation of the target compound with the protein was performed to output multiple binding poses. For example, a molecular docking simulation of 26 target compounds with the protein ERK2 resulted in the output of 257 binding poses. The same applies to other proteins.

[0038] Figure 7 shows an example of binding pose output from a molecular docking simulation. Figure 7 shows the proteins and the number of binding poses extracted from Figure 6. 257 binding poses are output for the protein ERK2, but among these, correct and incorrect binding poses are mixed, and it is impossible to tell at a glance whether these binding poses are correct or not. The same is true for other proteins.

[0039] Figure 8 shows an example of a binding pose labeled according to whether it is correct or incorrect. When the output binding pose is overlaid on the experimental structure, with the known experimental structure taken as the correct pose, it is clear that there is a mixture of correct binding poses that match the experimental structure and incorrect binding poses that do not match (are out of sync with) the experimental structure. Therefore, correct and incorrect binding poses are labeled based on the experimental structure. As shown in Figure 8, for the protein ERK2, of the 257 binding poses output, 12 are correct and 245 are incorrect. The same is true for other proteins.

[0040] Figure 9 shows an example of the evaluation results of a molecular dynamics simulation performed within a specified time. Using the labeling shown in Figure 8, a molecular dynamics simulation was performed within a specified time using the positional information of each atom in a protein-compound pair with a correct binding pose and the positional information of each atom in a protein-compound pair with an incorrect binding pose. The RMSD obtained by the molecular dynamics simulation within the specified time was plotted. For convenience, Figure 9 plots four binding poses, designated by symbols A to D.

[0041] Symbol D represents a binding pose with a small RMSD value and almost no change from the initial structure over time. Symbols A, B, and C represent binding poses with increasing RMSD values ​​over time. The black circles in the figure represent the maximum RMSD values ​​within 5 ns, the execution time of the molecular dynamics simulation.

[0042] If we create a histogram of the maximum RMSD values ​​within 5 nsec, with the horizontal axis representing the maximum RMSD value within 5 nsec and the vertical axis representing the number of poses, we can see that the maximum RMSD values ​​tend to be small for protein-compound pairs with correct binding poses, while the maximum RMSD values ​​tend to be large for protein-compound pairs with incorrect binding poses.

[0043] Figure 10 shows an example of a confusion matrix for determining whether a binding pose is correct or incorrect. Using the histogram shown in Figure 9, a predicted binding pose (P) is assigned to a pose with a maximum RMSD value smaller than a certain threshold, and a predicted binding pose (N) is assigned to a pose with a maximum RMSD value larger than the threshold. These are taken as the predicted labels (P, N) based on MD within a specified time. The docking results are taken as the true binding poses (P: correct binding pose, N: incorrect binding pose).

[0044] The true positive rate indicates whether the correct binding pose can be predicted as the correct binding pose, so for the confusion matrix on the left of Figure 10, the true positive rate = 85 / (85+133) = 0.39. Also, the false positive rate is the proportion of incorrect binding poses predicted as the correct binding pose, so for the confusion matrix on the left of Figure 10, the false positive rate = 63 / (63+3707) = 0.0167. Note that the smaller the false positive rate, the better the performance.

[0045] The threshold is changed and a confusion matrix is ​​created each time, and the true positive rate and false positive rate are calculated. For the confusion matrix on the right side of Figure 10, the true positive rate is 179 / (179+39)=0.82. Also, for the confusion matrix on the right side of Figure 10, the false positive rate is 613 / (613+3157)=0.163.

[0046] FIG. 11 is a diagram showing an example of ROC-AUC for determining whether a binding pose is correct or incorrect. As shown in FIG. 10, a ROC curve such as that shown in FIG. 11 is created by plotting the true positive rate and false positive rate calculated when the threshold is changed. ROC-AUC (Area Under the Curve) represents the area under the ROC curve. The maximum ROC-AUC is 1, and a larger value indicates higher discrimination ability. In the case of this embodiment, the ROC-AUC is 0.89, which shows that this is significantly more accurate than molecular docking simulation alone.

[0047] As described above, the binding information includes the binding pose of the compound molecule, and the control unit 51 can generate an evaluation index that can distinguish whether the binding pose is correct or incorrect.

[0048] By performing molecular dynamics simulation within a given time period, it is possible to distinguish between correct and incorrect binding poses, and to screen whether a compound molecule interacts with a target protein in a correct binding pose.

[0049] Next, it will be explained that RMSD can distinguish between the presence and absence of activity.

[0050] Figure 12 shows an example of target compounds labeled according to their activity values. A dataset of compound-protein pairs whose activity (activity information) is known through experiments is used. Compounds with activity values ​​less than 1 μM, i.e., highly active compounds, are classified as "active," while compounds with activity values ​​greater than 1 μM are classified as "inactive." Activity values ​​are values ​​that indicate the true magnitude of interaction between a compound and a protein, which can be obtained through experiments. As shown in Figure 12, for the protein ERK2, of the 64 target compounds, four are "active" and 60 are "inactive." The same is true for other proteins.

[0051] Figure 13 shows an example of a confusion matrix for determining whether or not there is activity. As in the case of Figure 10, a predicted label (P) of "active" is assigned to a molecule whose maximum RMSD value is smaller than a certain threshold, and a predicted label (N) of "inactive" is assigned to a molecule whose maximum RMSD value is larger than the threshold. These are defined as predicted labels (P, N) based on MD within a predetermined time. The actual presence or absence of activity is defined as P: "active" and N: "inactive."

[0052] The true positive rate indicates whether something that is actually "active" can be predicted as "active," so in the confusion matrix on the left of Figure 13, the true positive rate = 13 / (13 + 73) = 0.15. Also, the false positive rate is the proportion of "inactive" items predicted as "active," so in the confusion matrix on the left of Figure 13, the false positive rate = 13 / (9 + 690) = 0.0186.

[0053] The threshold is changed and a confusion matrix is ​​created each time, and the true positive rate and false positive rate are calculated. In the confusion matrix on the right side of Figure 13, the true positive rate = 69 / (69 + 17) = 0.80. Also, in the confusion matrix on the right side of Figure 13, the false positive rate = 69 / (137 + 562) = 0.099.

[0054] Fig. 14 is a diagram showing an example of ROC-AUC for determining the presence or absence of activity. As shown in Fig. 13, a ROC curve as shown in Fig. 14 is created by plotting the true positive rate and false positive rate calculated when the threshold is changed. ROC-AUC (Area Under the Curve) represents the area under the ROC curve. The maximum ROC-AUC is 1, and a larger value indicates higher discrimination ability. In the case of this embodiment, the ROC-AUC is 0.87, which shows that the accuracy is quite high.

[0055] As described above, the binding information includes the activity score of the compound molecule, and the control unit 51 can generate an evaluation index that can identify whether or not the compound is active.

[0056] By performing molecular dynamics simulation within a given time period, it is possible to distinguish whether or not there is activity, and to screen whether or not a compound molecule is a candidate compound that interacts with a protein molecule as a target protein.

[0057] The control unit 51 can identify the binding site where the compound molecule binds to the protein molecule based on the evaluation index. The correct binding pose and activity can be identified by molecular dynamics simulation within a predetermined time, so that the binding site where the compound molecule binds to the protein molecule can be identified.

[0058] Figure 15 shows an example of RMSD statistics. In Figure 15, the vertical axis represents ROC-AUC, and the horizontal axis represents the observation time of the molecular dynamics simulation from 0.01 to 100 nsec. Symbols A to D represent the statistical methods for RMSD, with symbol A representing the maximum RMSD value, symbol B representing the nth largest RMSD value, symbol C representing the average RMSD value, and symbol D representing the maximum average RMSD value per time t (nsec). Symbols A to D were used to verify the ability to distinguish between correct and incorrect binding poses. As shown in Figure 15, if the observation time (execution time) of the molecular dynamics simulation is at least 1 nsec, the ROC-AUC is 0.87, which is highly accurate. Although this depends on the computational power of the computer used for the simulation, sufficient accuracy can be expected with an observation time of 1 nsec, even if the computational power is not very high. Furthermore, if the computational power is not high, sufficient accuracy can be expected even with an observation time of less than 10 nsec. In other words, the predetermined time, which is the execution time of the molecular dynamics simulation, needs to be at least 10 nsec.

[0059] 16 is a diagram showing an example of a processing procedure of the information processing device 50. The control unit 51 acquires molecular coordinates of a protein (target protein) (S11), and acquires molecular coordinates of a compound to be screened for binding to the protein (S12). The control unit 51 executes a molecular docking simulation using the acquired molecular coordinates (S13).

[0060] The control unit 51 acquires the binding pose obtained by the molecular docking simulation (S14), and executes a molecular dynamics simulation within a predetermined time using the position information of each atom of the protein and compound (protein-compound pair) (S15). The control unit 51 acquires the evaluation index obtained by the molecular dynamics simulation (S16).

[0061] The control unit 51 screens compounds using the acquired evaluation index (S17). In screening compounds, for example, compounds with incorrect binding poses or no activity are excluded from the candidate compounds. The control unit 51 determines whether the target compound is a drug candidate compound (S18), and if it is a drug candidate compound (YES in S18), records the target compound as a drug candidate compound (S19), and performs the processing of step S20 described below.

[0062] If the target compound is not a drug candidate compound (NO in S18), the control unit 51 determines whether or not there is a compound to be screened (S20), and if there is a target compound (YES in S20), the control unit 51 continues the processing from step S12 onwards. If there is no target compound (NO in S20), the control unit 51 ends the processing. Note that the actual processing is not a step-by-step processing as shown in Figure 16, but is independent in each step, and the index is evaluated at the end.

[0063] According to this embodiment, compound-protein interactions can be captured using molecular dynamics (MD) simulations over a short period of time (within a specified time) on the order of nanoseconds, and a large number of target compounds can be screened accurately at low computational cost to obtain candidate drug compounds.

[0064] (Supplementary Note 1) The computer program causes a computer to execute a process of acquiring binding information between a protein molecule and a compound molecule obtained by molecular docking simulation, and generating, based on the acquired binding information, an evaluation index for screening whether or not the compound molecule is a candidate compound that interacts with the protein molecule as a target protein.

[0065] (Appendix 2) The computer program in Appendix 1 causes a computer to execute a process in which the binding information includes a predicted complex structure between the protein molecule and the compound molecule, and an evaluation index capable of identifying whether the predicted complex structure is correct or incorrect is generated.

[0066] (Supplementary Note 3) The computer program according to Supplementary Note 1 or Supplementary Note 2 causes a computer to execute a process in which the binding information includes an activity score corresponding to a predicted complex structure between the protein molecule and the compound molecule, and an evaluation index capable of identifying whether or not the compound is active is generated.

[0067] (Appendix 4) The computer program according to any one of Appendices 1 to 3 causes a computer to execute a process in which the binding information includes positional information of each atom of the protein molecule and the compound molecule, a molecular dynamics simulation is performed within a predetermined time to calculate a statistical value of the root mean square deviation of the positional information, and a change over time of the calculated statistical value within the predetermined time is generated as the evaluation index.

[0068] (Supplementary Note 5) In the computer program according to Supplementary Note 4, the predetermined time is on the order of nanoseconds.

[0069] (Supplementary Note 6) In the computer program according to Supplementary Note 4 or Supplementary Note 5, the predetermined time is less than 10 nsec.

[0070] (Supplementary Note 7) The computer program according to any one of Supplementary Note 1 to Supplementary Note 6 causes a computer to execute a process of identifying a binding site at which the compound molecule binds to the protein molecule based on the evaluation index.

[0071] (Appendix 8) The information processing device includes a control unit, which acquires binding information between a protein molecule and a compound molecule obtained by molecular docking simulation, and generates an evaluation index for screening whether or not the compound molecule is a candidate compound that interacts with the protein molecule as a target protein, based on the acquired binding information.

[0072] (Appendix 9) The information processing method acquires binding information between a protein molecule and a compound molecule obtained by molecular docking simulation, and generates, based on the acquired binding information, an evaluation index for screening whether the compound molecule is a candidate compound that interacts with the protein molecule as a target protein.

[0073] The matters described in each embodiment can be combined with each other. In addition, the independent claims and dependent claims described in the claims can be combined with each other in any combination, regardless of the reference format. Furthermore, although the claims use a format in which a claim references two or more other claims (multiple claim format), this is not limited to this format. A multiple claim (multi-multi claim) that references at least one other multiple claim may also be used. [Explanation of symbols]

[0074] 1. Communication Network 10 Molecular Docking Simulator 20 Molecular Dynamics Simulator 30 databases 31 Protein Data 32 Compound Data 50 Information processing equipment 51 Control section 52 Communications Department 53 Memory 54 Display section 55 Operation section 56 Memory section 57 Computer Programs 58 Recording medium reading unit

Claims

1. Binding information between protein molecules and compound molecules obtained by molecular docking simulation is acquired, generating an evaluation index for screening whether the compound molecule is a candidate compound that interacts with the protein molecule as a target protein based on the obtained binding information; A computer program that causes a computer to perform a process.

2. the binding information includes a predicted complex structure between the protein molecule and the compound molecule; generating an evaluation index capable of distinguishing whether the predicted complex structure is correct or incorrect; 2. The computer program according to claim 1, which causes a computer to execute a process.

3. the binding information includes an activity score corresponding to a predicted complex structure between the protein molecule and the compound molecule; generating an evaluation index capable of identifying the presence or absence of activity of the compound; 2. The computer program according to claim 1, which causes a computer to execute a process.

4. the binding information includes positional information of each atom of the protein molecule and the compound molecule, performing a molecular dynamics simulation within a predetermined time period to calculate a statistical value of the root mean square deviation of the position information; generating a temporal change in the calculated statistical value within the predetermined time period as the evaluation index; 4. A computer program according to claim 1, which causes a computer to execute a process.

5. The predetermined time is on the order of nanoseconds.

5. A computer program according to claim 4.

6. The predetermined time is less than 10 nsec.

5. A computer program according to claim 4.

7. identifying a binding site at which the compound molecule binds to the protein molecule based on the evaluation index; 4. A computer program according to claim 1, which causes a computer to execute a process.

8. A control unit is provided, The control unit Binding information between protein molecules and compound molecules obtained by molecular docking simulation is acquired, generating an evaluation index for screening whether the compound molecule is a candidate compound that interacts with the protein molecule as a target protein based on the obtained binding information; Information processing device.

9. Binding information between protein molecules and compound molecules obtained by molecular docking simulation is acquired, generating an evaluation index for screening whether the compound molecule is a candidate compound that interacts with the protein molecule as a target protein based on the obtained binding information; Information processing methods.

Images