Methods, devices and uses thereof for screening molecules

Abstract

The application relates to a method and device for screening molecules and application thereof. The method for screening molecules comprises the following steps: obtaining a first mapping relationship between M ligand molecules and N molecular structures; for at least part of the M ligand molecules, performing skeleton extraction on the structure information of the simplified molecular linear formula respectively to obtain O molecular skeletons; aggregating the O molecular skeletons to obtain P molecular skeleton classes; determining a second mapping relationship between the P molecular skeleton classes and the N molecular structures based on the first mapping relationship, so as to screen a molecule matched with a target receptor molecule from a simplified molecular linear formula set comprising the simplified molecular linear formula of the M ligand molecules based on the second mapping relationship. The scheme provided by the application can improve the accuracy and convenience of the user in screening molecules.

Description

Technical Field

[0001] The present application relates to the field of computational simulation technology, and in particular to a method and device for screening molecules and their application. Background Art

[0002] With the rapid development of computer technology and basic discipline theory, the computational efficiency and accuracy of molecular simulation have been greatly improved, making molecular simulation widely used in multiple disciplines. Among them, molecule screening is an important part of molecular simulation.

[0003] Related technologies can implement molecular screening based on thresholds for pre-set indicators. However, this may result in the elimination of some molecules that may be helpful for subsequent development. Summary of the Invention

[0004] In order to solve or partially solve the problems existing in the related art, the present application provides a method, device and application of screening molecules, which can reduce the probability of filtering out molecules that are helpful for subsequent development.

[0005] The first aspect of the present application provides a method for screening molecules, which includes: obtaining a first mapping relationship between simplified molecular linear formulas of M ligand molecules and N molecular structures, where the simplified molecular linear formulas of the M ligand molecules each have structural information, and M and N are integers greater than or equal to 1; for each molecule in at least some of the molecules in the simplified molecular linear formulas of the M ligand molecules, performing skeleton extraction on the structural information of the ligand molecule to obtain O molecular skeletons, where O is an integer greater than or equal to 1, and O is less than or equal to M; aggregating the O molecular skeletons to obtain P molecular skeleton classes, where P is an integer greater than or equal to 1, and P is less than or equal to O; determining a second mapping relationship between the P molecular skeleton classes and the N molecular structures based on the first mapping relationship, so as to screen ligand molecules that match the target receptor molecules based on the second mapping.

[0006] The second aspect of the present application provides a method for evaluating molecules, which includes: obtaining a simplified molecular linear formula of the molecule to be screened; determining the skeleton of the molecule to be screened based on the simplified molecular linear formula of the molecule to be screened; evaluating the molecule to be screened based on the skeleton of the molecule to be screened and multiple mapping relationships determined by the above method, the multiple mapping relationships including: at least one of the first mapping relationship to the sixth mapping relationship.

[0007] The third aspect of the present application provides a design method, which includes: displaying molecular screening results, which are screening results obtained according to the above method; and performing drug design or material design based on the molecular screening results.

[0008] The fourth aspect of the present application provides a device for screening molecules, comprising: a first mapping relationship obtaining module, configured to obtain a first mapping relationship between simplified molecular-input line-entry system (SMILES) of M ligand molecules and N molecular structures, wherein each of the SMILES of the M ligand molecules has structure information, M and N are integers greater than or equal to 1; a molecular skeleton extracting module, configured to, for each of at least part of the M ligand molecules, extract a skeleton from the structure information of the ligand molecule, to obtain O molecular skeletons, wherein O is an integer greater than or equal to 1 and less than or equal to M; a molecular skeleton aggregating module, configured to aggregate the O molecular skeletons to obtain P molecular skeleton classes, wherein P is an integer greater than or equal to 1 and less than or equal to O; and a second mapping relationship determining module, configured to determine a second mapping relationship between the P molecular skeleton classes and the N molecular structures based on the first mapping relationship, so as to screen a ligand molecule matching a target receptor molecule based on the second mapping relationship.

[0009] The fifth aspect of the present application provides a device for evaluating molecules. The device comprises: a SMILES obtaining module, configured to obtain a SMILES of a molecule to be screened; a skeleton obtaining module, configured to determine a skeleton of the molecule to be screened based on the SMILES of the molecule to be screened; and an evaluation module, configured to evaluate the molecule to be screened based on the skeleton of the molecule to be screened and a plurality of mapping relationships determined according to the device, wherein the plurality of mapping relationships comprises at least one of the first mapping relationship to the sixth mapping relationship.

[0010] The sixth aspect of the present application provides a design device, comprising: a screening result display module and a design module. The screening result display module is configured to display a molecule screening result, wherein the molecule screening result is obtained based on a screening result obtained by the device; and the design module is configured to perform drug design or material design based on the molecule screening result.

[0011] The seventh aspect of the present application provides an electronic device, comprising: a processor; and a memory having executable code stored thereon, wherein the executable code, when executed by the processor, causes the processor to perform the above method.

[0012] The eighth aspect of the present application further provides a computer-readable storage medium having executable code stored thereon, wherein the executable code, when executed by a processor of an electronic device, causes the processor to perform the above method.

[0013] The ninth aspect of the present application further provides a computer program product, comprising executable code, wherein the executable code, when executed by a processor, implements the above method.

[0014] The method and device for screening molecules and application thereof provided in the application determine the skeleton of a ligand molecule based on the molecular structure of the ligand molecule, cluster the skeletons of a plurality of ligand molecules to obtain skeleton classes, so as to realize the mapping relationship between the skeleton classes and the molecular structure, so that the molecular structure and other characteristics of a to-be-screened molecule can be predicted based on the skeleton of the to-be-screened molecule, and the accuracy and convenience of screening molecules are improved, so as to facilitate the recommendation of reasonable molecules into the synthesis and testing stage.

[0015] In addition, the technical solutions provided in the application can further determine the mapping relationship between the molecular structure and the structure class and / or the interaction class, so as to facilitate the user to screen molecules based on the mapping relationship in more dimensions.

[0016] In addition, the technical solutions provided in the application can verify whether the interaction in the interaction class is stable based on the result of the kinetic simulation, so as to facilitate the user to screen molecules based on whether the interaction is stable.

[0017] It should be understood that the foregoing general description and the following detailed description are only exemplary and explanatory, and cannot limit the application. BRIEF DESCRIPTION OF DRAWINGS

[0018] The above and other objects, features and advantages of the present application will become more apparent from the following detailed description when taken in conjunction with the accompanying drawings, in which like reference characters refer to like parts throughout the several views, and in which the exemplary embodiments of the present application are shown.

[0019] Figure 1 A process diagram of screening molecules according to an embodiment of the application is schematically shown;

[0020] Figure 2 An exemplary system architecture to which the method and device for screening molecules and application thereof according to an embodiment of the application can be applied is schematically shown;

[0021] Figure 3 A flowchart of a method for screening molecules according to an embodiment of the application is schematically shown;

[0022] Figures 4 to 6 A process diagram of extracting a molecular skeleton according to an embodiment of the application is schematically shown;

[0023] Figure 7 A schematic diagram of a skeleton graph according to an embodiment of the application is schematically shown;

[0024] Figure 8A A structure diagram of a skeleton according to an embodiment of the application is schematically shown;

[0025] Figure 8Bschematically illustrates a structural schematic diagram of another skeleton according to an embodiment of the present application;

[0026] Figure 9 schematically illustrates a flow chart of a molecule evaluation method according to an embodiment of the present application;

[0027] Figure 10 schematically illustrates a flow chart of a design method according to an embodiment of the present application;

[0028] Figure 11 schematically illustrates a block diagram of a molecule screening device according to an embodiment of the present application;

[0029] Figure 12 schematically illustrates a block diagram of a molecule evaluation device according to an embodiment of the present application;

[0030] Figure 13 schematically illustrates a block diagram of a design device according to an embodiment of the present application;

[0031] Figure 14 schematically illustrates a block diagram of an electronic device implementing a molecule screening method according to an embodiment of the present application. DETAILED DESCRIPTION

[0032] Embodiments of the present application will be described in more detail by referring to the drawings. Although the embodiments of the present application are shown in the drawings, it should be understood that the present application can be implemented in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided so that the present application will be thorough and complete, and will fully convey the scope of the present application to those skilled in the art.

[0033] The terms used in the present application are merely used to describe particular embodiments, and are not intended to limit the present application. The terms "include", "comprise" and the like used herein indicate the presence of the features, steps, operations and / or components, but do not exclude the presence or addition of one or more other features, steps, operations or components.

[0034] All terms used herein (including technical and scientific terms) have meanings commonly understood by one of ordinary skill in the art, unless otherwise defined. It should be noted that the terms used herein should be interpreted as having meanings consistent with the context of the present specification, and should not be interpreted in an idealized or overly formal manner.

[0035] It should be understood that, although the terms "first", "second", "third", etc. can be used herein to describe various information, the information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, the first information can also be referred to as the second information, and similarly, the second information can also be referred to as the first information without departing from the scope of the present application. Therefore, the features defined with "first", "second" can explicitly or implicitly include one or more of the features. In the description of the present application, the meaning of "a plurality of" is two or more, unless otherwise specifically limited.

[0036] Before the technical solutions of the present application are described, some technical terms in the field related to the present application are explained.

[0037] Molecular docking is a molecular simulation method that mainly analyzes the characteristics and interactions of receptors and ligands through electric field forces, thereby facilitating the prediction of the binding mode of the receptor and the ligand.

[0038] Molecular simulation refers to the use of theoretical methods and computer technology to simulate the structure and physical and chemical properties of molecules or molecular systems.

[0039] In order to solve the problems existing in the related art, virtual screening based on molecular structure can be applied to the early stage of material development, such as drug development. The role of virtual screening is to screen potential ligand molecules (such as drug molecules) that can bind to target receptor molecules (such as proteins) from a large-scale (for example, the number of molecules > 10 8

[0040] ​In the related art, a virtual screening algorithm can be used to score the interaction between a drug molecule and a protein, and a molecule with a high score means that it has more potential to become a candidate drug molecule to enter the next stage of development. Such software is usually referred to as molecular docking software. For example, molecular docking software includes but is not limited to AutoDockVina, ICM LeDock, rDock, UCSF DOCK, etc., and commercial software includes Glide, LigandFit, GOLD, MOE Dock, etc. It should be noted that the screening method based on molecular structure has become one of the paradigms of early small molecule drug research in the related art, and the compound library available for screening, such as Enamine Real, Labnetworkx, etc. The compounds contained in these compound libraries are over one hundred million, and the rapid development of cloud computing makes large-scale molecular screening calculation possible. However, large-scale compound screening brings challenges to post-processing, especially the selection of promising compounds that meet the specific target binding pocket. The drug molecule screening process in the related art scores the molecules in the screened compound library through one or more such molecular docking software, sets a relatively acceptable threshold. Molecules higher than the threshold are retained for use in subsequent development processes.

[0041] Specifically, first, a three-dimensional (3D) structure of a molecule to be screened is established and energy optimization is performed. Then, hydrogen atoms are added to the 3D structure, and a force field and atomic charges of the molecule to be screened are added respectively. Next, a template target receptor molecule is generated with a probe of a preset radius. Then, a reasonable binding site of the molecular surface of the target receptor molecule is determined. Energy scoring and evaluation are performed for each reasonable binding site. Specifically, semi-flexible docking or the like can be used to generate a specified number of different conformational orientations, obtain the electrostatic and van der Waals interaction between the molecule to be screened and the binding site, and calculate the score for the molecule to be screened. The score is compared with a preset threshold to determine whether the molecule to be screened is retained for subsequent development processes.

[0042] However, the AUC of such scoring algorithms or software is less than 80% on the test set, resulting in that defining a filtering condition by a certain threshold may cause molecules that can be helpful for subsequent development to be filtered out too early. Among them, AUC (area under curve) is defined as the area surrounded by the receiver operating characteristic curve (ROC) and the coordinate axis, and the higher the AUC is close to 1.0, the higher the authenticity of the detection method is.

[0043] For example, the applicant found that the docking-based algorithm in the related art can also perform molecular screening from a large-scale compound library by selecting a specific docking scoring threshold, but the overall accuracy (such as AUC) of positive molecular screening is often related to the accuracy of the docking algorithm and the threshold selected by the user. The screening method may miss potential lead compounds with optimization space in the early drug development application because the docking algorithm scoring is not high (lower than the threshold selected by the user or the docking algorithm is not accurate).

[0044] In summary, how to seek an effective molecular evaluation method other than the method of only considering the docking score value has become a problem to be improved in the process of virtual screening.

[0045] In addition, in the early drug screening stage, especially for the early drug development of the first drug (First-in-class, FIC for short) of the same type, the developer expects the evaluation result of the molecule to help the strategy of recommending the molecule to synthesis and testing. Structure-based drug design usually establishes a hypothetical correlation between small molecule structure information and protein binding mode, and recommends synthesis, testing, and verification of the hypothesis on this basis to facilitate subsequent improvement. However, the docking software is more evaluated from the perspective of small molecule and protein structure, and how to aggregate the docking results of many small molecules, aggregate important structural, skeletal, and binding mode difference information, and use it as a basis to assist in recommending molecules to the subsequent synthesis and testing process is particularly important.

[0046] In the related art, evaluation algorithms or software such as quantitative structure-activity relationship (QSAR for short) can be used to evaluate drug molecules and consider factors related to the binding of the molecule to be screened to the protein. However, the evaluation result does not directly correlate with the three-dimensional spatial structure of the small molecule occupying the protein pocket, and it is difficult to establish a hypothetical correlation between the small molecule structure and the binding mode, thereby limiting the role of assisting the recommendation of synthesis and testing.

[0047] The present application aims to provide a molecular evaluation method for a virtual screening process, which aggregates information of at least one of the molecular structure, the skeleton, and the binding mode of the molecules that successfully dock to facilitate the recommendation of reasonable molecules into the synthesis and testing stage.

[0048] The technical solution of the present application can adapt to the increase of positive compound scale, and provide a molecular screening method more in line with drug development experience. The technical solution of the present application can project complex high-latitude small molecule protein spatial structure information to the skeleton, shape and binding mode in hierarchical manner. And the representative molecules can be selected by clustering algorithm based on this. In actual situation, the drug chemist can select the appropriate number of representative molecules according to the objective factors such as the stage of the specific project, risk and budget to regulate the clustering and screening algorithm of the method, so as to realize the auxiliary decision-making ability from large-scale virtual screening to pipeline molecule recommendation. It should be noted that the technical solution of the present application is more suitable for the input (molecule library), and can be applied to any size of molecule library.

[0049] In addition, the embodiment of the present application breaks out of the method of screening molecules with score value as the only reference value, and returns to the structure-based rational design. The superior skeleton molecules are selected from the rationality of structure to provide downstream molecule optimization selection.

[0050] The following will be described in detail by Figures 1 to 13 The method for screening molecules, the device and the application thereof according to the embodiment of the present application will be described in detail.

[0051] Figure 1 The process schematic diagram of screening molecules according to the embodiment of the present application is schematically shown.

[0052] Referring to Figure 1 In the embodiment, at least one of the skeleton class, structure class and related action class of the ligand molecules which are successfully docked is associated with the molecular structure and the simplified molecular linear formula, so that the user can screen the screening molecules with the aid of these knowledge, and the molecular screening accuracy is improved.

[0053] Figure 2 An exemplary system architecture to which the method for screening molecules, the device and the application thereof according to the embodiment of the present application can be applied is schematically shown.

[0054] It should be noted that Figure 2 The system architecture shown is only an example of the system architecture to which the embodiment of the present application can be applied, to help those skilled in the art understand the technical content of the present application, but does not mean that the embodiment of the present application cannot be applied to other devices, systems, environments or scenarios.

[0055] Referring to Figure 2 According to the embodiment, the system architecture 200 can include terminal devices 201, 202 and 203, a network 204 and a server 205. The network 204 is used as a medium to provide a communication link between the terminal devices 201, 202, 203 and the server 205. The network 204 can include various connection types, such as wired, wireless communication links or optical fiber cables, etc.

[0056] Users can use terminal devices 201, 202, and 203 to interact with other terminal devices and server 205 via network 204 to receive or send information, such as sending a request for a molecule's three-dimensional molecular structure, a request for screening a molecule, and receiving screening results, a molecule's three-dimensional molecular structure, etc. Terminal devices 201, 202, and 203 can be installed with various communication client applications, such as web browser applications, drug development applications, database applications, search applications, instant messaging tools, email clients, social platform software, and the like.

[0057] The terminal devices 201 , 202 , and 203 include, but are not limited to, smart desktop computers, tablet computers, laptop computers, and other electronic devices that can support functions such as surfing the Internet and displaying images.

[0058] The server 205 may receive a molecule three-dimensional molecular structure request and send the molecule three-dimensional molecular structure information to the terminal devices 201, 202, and 203. For example, the server 205 may be a background management server, a server cluster, or the like.

[0059] It should be noted that the number of terminal devices, networks, and servers is only illustrative. Any number of terminal devices, networks, and clouds may be provided as required.

[0060] Figure 3 The figure schematically shows a flow chart of a method for screening molecules according to an embodiment of the present application.

[0061] See also Figure 3 This embodiment provides a method for screening molecules, which includes operations S310 to S340, specifically as follows:

[0062] In operation S310 , a first mapping relationship between simplified molecular linear formulas of M ligand molecules and N molecular structures is obtained, where each simplified molecular linear formula of the M ligand molecules has structural information, and M and N are integers greater than or equal to 1.

[0063] In this embodiment, the ligand molecule is the molecule corresponding to the receptor molecule. For example, if small molecule A and macromolecule B are successfully docked, the small molecule can be called a ligand molecule. The structural information can be the structural information contained in the simplified molecular linear formula.

[0064] Molecular docking is a method for drug design that uses the characteristics of receptor molecules (such as protein molecules) and the interaction between receptor molecules and ligand molecules (such as drug molecules). Molecular docking studies the interaction between molecules (such as ligand molecules and receptor molecules) and predicts their binding mode and affinity.

[0065] The drug molecule and the protein macromolecule in the body can have a key and lock recognition relationship. The recognition relationship can depend on the spatial matching and energy matching of the two. For example, taking a virus as an example, the RNA of the virus depends on a certain RNA polymerase protein. It has been proved that a certain part of the RNA polymerase protein is the target of small molecule drugs. Molecular docking can be used to predict the binding activity of multiple small molecules and the target, so as to predict whether these small molecules have the potential to become candidate drugs.

[0066] A simplified molecular linear formula can express the structure of a molecule in text. Specifically, the simplified molecular linear formula can comply with the Simplified Molecular-Input Line-Entry System (SMILES). The SMILES method represents the structure of a molecule as text. By converting the structure information into text information, the text (encoded string) is used as input in the machine learning input pipeline. In this way, natural language processing (NLP) related algorithms can be used for drug development.

[0067] There can be multiple corresponding molecular structures for one simplified molecular linear formula. The first mapping relationship can be as follows. The simplified molecular linear formula corresponds to the smiles column, and the molecular structure corresponds to the sdf_index column. The first mapping relationship can be represented as: sdf_index, similes.

[0068] For example, the first mapping relationship 1:

[0069] protein_ligand_02538_energy_0_split_0_pose_8,CCl(C)CC(C[NH2+]CC(=O)

[0070] The first mapping relationship 2:

[0071] protein_ligand_02538_energy_0_split_0_pose_3,CCl(C)CC(C[NH2+]CC(=O)

[0072] The first mapping relationship 3:

[0073] protein_ligand_02538_energy_0_split_0_pose_5,CCl(C)CC(C[NH2+]CC(=O)

[0074] In the first mapping relationship 1, the first mapping relationship 2, and the first mapping relationship 3, the sdf_index column is before the “,”, and the smiles column is after the “,”.

[0075] In one embodiment, the molecular structures of a series of compound molecules (structures after successful docking) are obtained from the virtual screening process, and are stored in a file in.sdf format, wherein only the structures of the ligand small molecules can be stored. In addition, there is a.csv file storing the mapping relationship between the SMILES of each molecule and the file (.sdf) of the corresponding molecular structure, i.e., a first mapping relationship. The SMILES of one molecule can correspond to multiple molecular structures.

[0076] In operation S320, for at least part of the simplified molecular linear formulas of the M ligand molecules, the structural information of the simplified molecular linear formula is respectively subjected to skeleton extraction to obtain O molecular skeletons, O is an integer greater than or equal to 1, and O is less than or equal to M.

[0077] In some embodiments, various skeleton extraction methods can be used to extract the skeleton from the simplified molecular linear formula. For example, a skeleton extraction algorithm is performed on the SMILES of each molecule to obtain the Bemis-Murcko skeleton of each molecule.

[0078] Figures 4 to 6 A process diagram for extracting a molecular skeleton according to an embodiment of the present application is schematically shown.

[0079] Referring to Figure 4 , the SMILES 1 can be converted into the molecular structure diagram shown in Figure 4 . The conversion method between the SMILES and the molecular structure diagram can employ various related technologies, which are not described in detail here.

[0080] Referring to Figure 5 , the nitrogen atom (N) and the oxygen atom (O) in Figure 4 are replaced, such as being replaced by a carbon atom (C) or a hydrogen atom (H). It should be noted that the element symbol identifiers for the carbon atom (C) and the hydrogen atom (H) are omitted in Figure 5 . In addition, the element symbol identifiers in Figure 4 may also be deleted. Then, the double bond in Figure 4 is replaced by a single bond. Through the above operations, the structure shown in Figure 4 can be obtained. The above is only an exemplary illustration and should not be construed as a limitation on the present application. Figure 5

[0081] Referring to Figure 6 , the nitrogen atom (N) and the oxygen atom (O) in Figure 5 are replaced, such as being replaced by a carbon atom (C) or a hydrogen atom (H). It should be noted that the element symbol identifiers for the carbon atom (C) and the hydrogen atom (H) are omitted in Figure 6 ​The branched chain or the hanging bond in the SMILES formula is removed. The skeleton corresponding to the SMILES formula can be obtained by the above operation. For example, the skeleton corresponding to the SMILES formula can be obtained by Bemis-Murcko framework.

[0082] In operation S330, the O molecular skeletons are aggregated to obtain P molecular skeleton classes, P is an integer greater than or equal to 1, and P is less than or equal to O.

[0083] In this embodiment, a plurality of skeleton classes can be obtained by the aggregation operation on the plurality of molecular skeletons. For example, the same skeleton is classified into a class.

[0084] In some embodiments, after obtaining the P molecular skeleton classes, the method can further include the following operation: performing intra-class merging on the P molecular skeleton classes to obtain a multi-level molecular skeleton set, wherein a parent molecular skeleton in the multi-level molecular skeleton set corresponds to at least one child molecular skeleton, a bottom-level molecular skeleton corresponds to at least one ligand molecule, and the skeleton structure of the child molecular skeleton is more complex than the skeleton structure of the parent molecular skeleton.

[0085] By intra-class merging, at least part of the ligand molecules can be integrated into a table, so that the parent-child relationship between various skeletons is more clear, so that the user can select a possible useful skeleton or molecule based on the parent-child relationship.

[0086] In some embodiments, after obtaining the multi-level molecular skeleton set, the method can further include the following operation: generating a skeleton graph, wherein the skeleton graph includes a plurality of nodes, a non-terminal node in the plurality of nodes represents at least part of the molecular skeletons in the multi-level molecular skeleton set, a terminal node in the plurality of nodes represents a molecular cluster in the M ligand molecules including the skeleton corresponding to the terminal node, and a parent node in the plurality of nodes corresponds to at least one child node.

[0087] Figure 7 A schematic diagram of a skeleton graph according to an embodiment of the present application is schematically shown.

[0088] Referring to Figure 7 , Figure 7 Each node in the skeleton graph can represent a skeleton class, a parent node corresponds to at least one child node, and a bottom-level node can correspond to a molecular formula of a specific ligand molecule. As shown in Figure 7 The two nodes enclosed by the two dashed circles on the left side in the skeleton graph are the molecular formulas of two ligand molecules. The two ligand molecules have the same skeleton class. However, the skeleton class in which the node enclosed by the upper dashed circle is located is a parent-child relationship with the skeleton class in which the node enclosed by the lower dashed circle is located.

[0089] For example, the graph allows for intuitive visualization of the relationships between molecules and backbones, with each terminal leaf node representing a cluster of molecules sharing the same backbone. By numbering each node, we can obtain the backbone number based on backbone clustering. For example, the root node is 1, and each node is numbered sequentially using Arabic numerals, such as 1, 2, 3, 4, and so on, as the graph is traversed.

[0090] also, Figure 7 Each node in the diagram can have a fill color or fill pattern, for example. Taking fill color as an example, the depth of the fill color can indicate the average score of all molecules contained in the current skeleton node. For example, if the score represents activity, then the darker the node, the higher the activity value of the molecule contained in its skeleton, that is, the higher the match between the ligand molecule corresponding to the node and the target receptor molecule. This makes it easier for users to intuitively see from the skeleton diagram: which skeleton contains the molecule with the higher activity and the higher probability of becoming the ligand molecule corresponding to the target receptor molecule.

[0091] Figure 8A A schematic structural diagram of a skeleton according to an embodiment of the present application is shown schematically. Figure 8B A schematic structural diagram of another skeleton according to an embodiment of the present application is shown schematically.

[0092] See also Figure 8A and Figure 8B , Figure 8A This is the scaffold for the ligand molecule identified as MOL0436. Figure 8B These are the skeletons of the ligand molecule identified as MOL0049. The middle parts of the two skeletons framed by dotted circles are the same, and the two skeletons have the same parent skeleton.

[0093] In operation S340, a second mapping relationship between the P molecular scaffold classes and the N molecular structures is determined based on the first mapping relationship, so as to screen molecules matching the target receptor molecules from the simplified molecular linear formula set including the simplified molecular linear formulas of the M ligand molecules based on the second mapping relationship.

[0094] In this embodiment, by constructing the second mapping relationship, it is convenient for the user to perform molecular screening based at least on the second mapping relationship.

[0095] In some embodiments, in order to facilitate users to perform molecular screening from more dimensions, the above method may further include the following operations.

[0096] First, for each of at least some of the P molecular skeleton classes, obtain the molecular structures corresponding to at least some of the ligand molecules corresponding to the skeleton class. Figure 7 , you can get all the molecular structures corresponding to a certain node.

[0097] Then, volume differences between the at least part of the ligand molecules are determined based on respective molecular structures of the at least part of the ligand molecules.

[0098] Next, the at least part of the ligand molecules corresponding to the skeleton class are clustered based on the volume differences, to obtain a plurality of structure classes. For example, ligand molecules with small volume differences are clustered into one class.

[0099] Then, a third mapping relationship between the plurality of structure classes and the molecular structures is determined based on the first mapping relationship.

[0100] The determining of the volume differences between the at least part of the ligand molecules based on respective molecular structures of the at least part of the ligand molecules can include the following operations.

[0101] First, a pocket region of the target receptor molecule is divided into a grid.

[0102] Then, an occupation proportion of the molecular structure corresponding to each of the at least part of the ligand molecules to the grid is determined.

[0103] Next, a grid space occupation vector of the molecular structure corresponding to each of the at least part of the ligand molecules is constructed based on the occupation proportion.

[0104] Then, the volume differences between the at least part of the ligand molecules are determined based on the grid space occupation vectors of the molecular structures corresponding to the at least part of the ligand molecules.

[0105] For example, the determining of the volume differences between the at least part of the ligand molecules based on the grid space occupation vectors of the molecular structures corresponding to the at least part of the ligand molecules can include the following operations.

[0106] First, a distance between the grid space occupation vectors of the molecular structures corresponding to two ligand molecules is determined.

[0107] Then, the volume difference of the space occupied by the molecular structures corresponding to the two ligand molecules is determined based on the distance between the grid space occupation vectors.

[0108] The above operations are repeated until the volume differences between the molecular structures corresponding to any two ligand molecules in the at least part of the ligand molecules are determined.

[0109] In one embodiment, first, three-dimensional molecular structures of all molecules in a skeleton class are extracted, such as from a.sdf file. Then, spatial occupancy volume differences between three-dimensional molecular structures of each pair of molecules are calculated. Since the molecular structures are from three-dimensional molecular structures after docking, structure-based alignment and translation change operations are not necessary here. After obtaining the spatial occupancy volume differences between each pair of molecules, clustering analysis based on three-dimensional shape differences can be performed. Each class represents molecules with similar spatial occupancy volumes. For example, molecules with spatial distances less than a preset distance threshold can be clustered into the same class, and the preset distance threshold can be taken as ~4 angstroms, or and the like. The spatial occupancy volume difference between three-dimensional molecular structures of each pair of molecules can be determined by dividing the pocket region of the molecular structure of the receptor molecule (such as a protein structure) into an equidistant grid, determining the occupancy of the molecular structure to the grid, constructing a grid spatial occupancy vector, and then calculating the distance (such as Tanimoto Distance, Euclidean Distance, etc.) between two grid spatial occupancy vectors to obtain the spatial occupancy volume difference between two molecules.

[0110] In this embodiment, according to the theoretical basis of the “lock and key principle” and “induced fit”, it can be determined that the ligand molecule can be correctly combined in the protein pocket. The ligand molecule should be in shape and electrostatic distribution with the pocket (complementary). The structure matching degree between the molecule and the target receptor molecule can be determined by the above method.

[0111] In some embodiments, after obtaining a plurality of structure classes, the above method can further include the following operations to further analyze the mapping relationship between the structure class and / or skeleton class and the molecular structure.

[0112] First, for each class of at least part of the structure classes, obtain the first interaction feature between each of the at least part of the molecular structures corresponding to the structure class and the target receptor molecule, and / or for each class of at least part of the P molecular skeleton classes, obtain the second interaction feature between each of the at least part of the molecular structures corresponding to the skeleton class and the target receptor molecule.

[0113] Then, determine the first interaction difference between the first interaction features of the at least part of the molecular structures, and / or determine the second interaction difference between the second interaction features of the at least part of the ligand molecules.

[0114] Next, the first interaction difference is clustered to obtain a plurality of first interaction classes, and / or the second interaction difference is clustered to obtain a plurality of second interaction classes.

[0115] Then, a fourth mapping relationship between the plurality of first interaction classes and the molecular structure is determined based on at least the first mapping relationship, and / or a fifth mapping relationship between the plurality of second interaction classes and the molecular structure is determined based on the first mapping relationship.

[0116] In the embodiment, clustering analysis based on the molecular-protein binding mode fingerprint can be implemented. The molecular-protein binding mode fingerprint is a code for characterizing the interaction type between a small molecule and a protein molecule based on structure. For example, the interaction fingerprint includes, but is not limited to, at least one of the interaction type, the interaction site atom serial number, and the interaction site of the protein amino acid residue.

[0117] It should be noted that after obtaining the skeleton classes, the process of determining the interaction classes can be performed without obtaining the structure classes.

[0118] For example, after obtaining P molecular skeleton classes, the method can further include the following operations.

[0119] First, for each of at least part of the P molecular skeleton classes, the second interaction features between each of the at least part of the ligand molecules corresponding to the skeleton class and the target receptor molecule are obtained.

[0120] Then, the second interaction differences between the second interaction features of the at least part of the ligand molecules are determined.

[0121] Next, the second interaction differences are clustered to obtain a plurality of second interaction classes.

[0122] Then, a fifth mapping relationship between the plurality of second interaction classes and the molecular structure is determined based on at least the first mapping relationship.

[0123] In some embodiments, determining the second interaction differences between the second interaction features of the at least part of the ligand molecules can include the following operations.

[0124] First, the interaction feature vectors corresponding to the second interaction features are determined.

[0125] Then, the following operations are repeated until the interaction differences between any two of the second interaction features of the at least part of the ligand molecules are determined. For example, first, the distances between the interaction feature vectors corresponding to the second interaction features are determined, and then the second interaction differences of the two ligand molecules are determined based on the distances between the interaction feature vectors corresponding to the second interaction features.

[0126] In one embodiment, the encoding information of the molecular-protein molecule binding mode fingerprint includes interaction types (such as hydrogen bond donor molecules and acceptor molecules, π-π interactions, etc.), atom numbers of the interaction sites of the molecules, and the interaction sites of the protein amino acid residues. Through this information, the structure-based interaction formed by the molecules and the protein molecules can be quickly identified. Each molecule can form multiple such interactions with the protein molecules, and each molecule can extract multiple interaction fingerprints (interaction fingerprints are obtained by vectorizing the interaction characteristics of the molecules, and a 1 x n-dimensional vector is obtained. The distance between interaction fingerprints can be calculated by the Yubin distance method, etc.) from the molecular structure after docking. According to the interaction fingerprints, the fingerprint feature vector of the molecular structure can be constructed. For example, for all molecular three-dimensional molecular structure information of a certain skeleton class and / or a certain structure class, fingerprint extraction can be performed, and the fingerprint distance between two fingerprints can be calculated, and then fingerprint-based clustering analysis can be performed. Molecules in the same class should have similar skeletons, shapes, and / or binding modes. Fingerprint clustering can be regarded as an unsupervised clustering.

[0127] In some embodiments, considering that the interaction can be divided into stable interaction and unstable interaction, in this embodiment, the stability characteristics of the interaction can be further analyzed.

[0128] Specifically, after obtaining the plurality of first interaction classes, or after obtaining the plurality of second interaction classes, the above method can further include the following operations.

[0129] First, for any one of the first interaction classes or the second interaction classes, a representative molecule of the current class is determined. For example, the representative molecule can be a molecule corresponding to the class center of a certain class, etc.

[0130] Then, molecular dynamics simulation is performed on the representative molecule of the current class to obtain the stability characteristics of the representative molecule. The stability characteristics can include: stable and unstable.

[0131] Next, based on the stability characteristics of the representative molecule, a sixth mapping relationship between the stability characteristics and the first interaction class or the second interaction class is determined.

[0132] In one embodiment, representative molecules in each cluster can be determined for performing a dynamics simulation. The purpose is to verify whether the interaction is still stable in the result of the dynamics simulation, and if there is an unstable interaction under the dynamics model, it should be identified in the final result. Specifically, a 50ns dynamics simulation can be performed on the complex structure of the successfully docked ligand molecule and the receptor molecule, and a trajectory file is collected by sampling. Based on the trajectory file, it is analyzed whether the interaction between the extracted ligand molecule and the protein molecule can be continuously observed in the stable state of sampling. If it can be continuously observed, it means that the interaction formed by the representative molecule in the simulation is still stable and observable. The representative molecule can be a cluster center molecule, which is a point with the most balanced distance between the object and the remaining objects in the class. The cluster center can only have one, and can be directly obtained by a clustering algorithm.

[0133] In some embodiments, in order to facilitate the user to view the above-mentioned various mapping relationships, the above-mentioned method can further include the following operations.

[0134] The simplified molecular linear formula, the molecular structure of the ligand molecule, and at least one of the following: the molecular skeleton class, the structure class, the first interaction class, or the second interaction class are stored in association, to obtain a mapping table.

[0135] Specifically, the information obtained by the above-mentioned operations can be summarized to form a summary table of "molecule (simplified molecular linear formula)-molecular structure-skeleton class-structure class-interaction class-stability", so as to facilitate the drug designer to select appropriate molecules as hypotheses according to the skeleton, interaction class (binding mode) of interest, and then perform subsequent synthesis and test verification.

[0136] In one embodiment, 32k mapping relationships and 4k different simplified molecular linear formulas are recorded. The.csv relationship file can be loaded by using the Python Pandas library, and all data in the "smiles" column of the table are obtained, that is, all the simplified molecular linear formulas. The Bemis-Murcko skeleton of each simplified molecular linear formula can be extracted by the above-mentioned related content, and the same skeleton can be combined. In addition, the related node information can be represented and stored in the form of a skeleton graph to realize visualization. For example, the skeleton graph can be stored by using the Python Networkx library, and a node graph as shown in Figure 7 can be drawn.

[0137] From Figure 7The relationship between the simplified molecular linear formula and the scaffold is intuitively visible in the graph. Each underlying leaf node represents a cluster of molecules with the same terminal scaffold. By numbering each node in the scaffold graph, we obtain the scaffold number based on scaffold clustering. This number is also persisted in the input .csv relational table, labeled "scaffold_cluster."

[0138] Next, we conduct structure-based cluster analysis. Figure 7 For any node in the skeleton, extract the simplified linear molecular formula and the corresponding molecular structure .sdf file from the corresponding cluster. Using methods such as RDKit's Shape Protrude Distance, calculate the spatial volume difference between each pair of molecular structures, which serves as the distance between them. Then, cluster all molecular structures using algorithms such as the DBSCAN algorithm from the scikit-learn machine learning development package. Determine the molecular structure classes for all molecules within each skeleton class. Sequentially number the class classes and update them into the input .csv relational table. For example, they can be labeled "shape_cluster."

[0139] Cluster analysis based on interaction fingerprints is similar to cluster analysis based on structure. .sdf files of all molecular structures under each skeleton class are obtained. At the same time, the protein structure file .pdb used for docking needs to be obtained. The fingerprint information between the ligand molecule and the protein molecule is extracted by combining the pattern fingerprint recognition algorithm, as shown below:

[0140] 6UYB_500ns_frame_ligand_009433_energy_2_isomer_0_split_0_pose_0[0,0,0,0, 0,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,0,1,0,1,0,0,0]

[0141] Each digit in the brackets represents a one-dimensional feature. The value of the feature can be obtained by one-hot encoding. For example,

[0142] The encoded information includes the interaction type (such as hydrogen bond donor acceptor, π-π interaction, etc.), the atomic number of the small molecule's action site and the action site of the protein amino acid residue.

[0143] For example, if the current small molecule and protein have a hydrogen bond interaction on the CYS260 amino acid, and the small molecule is a hydrogen bond acceptor, the corresponding interaction is coded as [CYS260_HB_Acceptor, ...].

[0144] Thus, each small molecule will form multiple interactions with the protein, i.e. each small molecule has a list of interactions as above, such as [CYS260_HB_Acceptor,...].

[0145] Collecting all the interactions, an interaction feature vector is built in an arbitrary determined order, then for any small molecule, the existing interactions are coded as 1 and the non-existing ones are coded as 0. In this way, a feature vector of the same length can be built for all molecules, such as [0, 0, 1, 0,...].

[0146] Next, after encoding the interaction features to get the feature vector, clustering based on the distance between fingerprints can be performed by a clustering algorithm. The cluster number of the fingerprint-based clustering is stored in the input.csv relationship table, which can be marked as "ifp cluster".

[0147] For the representative molecule (e.g. cluster center) in each fingerprint cluster, its small molecule structure.sdf file and corresponding protein structure.pdb file are extracted, and a specified duration (e.g. 100 ns) of dynamics simulation is performed via GROMACS, and a number of frames are sampled from the simulation trajectory to identify the interaction binding mode, which is used to verify the effectiveness of the interaction fingerprint. If the interactions indicated by the fingerprint still exist stably in the simulation, it is marked as 1; if not, it is marked as 0. This is recorded in the input.csv relationship table, which can be marked as "ifp_valid".

[0148] An entry in the.csv table after summarizing the above results can be as shown in Table 1.

[0149] Table 1

[0150]

[0151] Another aspect of the present application also provides a method for evaluating molecules.

[0152] Figure 9 A flowchart of a method for evaluating molecules according to an embodiment of the present application is schematically shown.

[0153] Referring to Figure 9 The user can use the mapping table in the following way. The above method can further include operation S910 to operation S930.

[0154] In operation S910, a simplified molecular linear formula of a molecule to be screened is obtained. For example, the user can input the simplified molecular linear formula on the terminal device, and the terminal device can also send the simplified molecular linear formula to the cloud.

[0155] At operation S920, a skeleton of the molecule to be screened is determined based on the simplified molecular linear expression of the molecule to be screened. The skeleton corresponding to the simplified molecular linear expression can be generated locally or in the cloud.

[0156] At operation S930, the molecule to be screened is evaluated based on the skeleton of the molecule to be screened and the plurality of mapping relationships determined by the method described above. The plurality of mapping relationships can be stored in a mapping table, including but not limited to at least one of the first mapping relationship to the sixth mapping relationship. For example, the entry corresponding to the molecule to be screened in the mapping table can be determined by skeleton matching or the like.

[0157] For another example, a user can select a skeleton, a shape, and an interaction cluster of interest, and then select a representative molecule in the cluster to enter a subsequent synthesis and test phase, and verify whether the skeleton, the shape, and the interaction are helpful for drug design of the target protein according to the results of the synthesis and test. Meanwhile, a batch of molecules can be combined according to different skeleton, shape, and interaction clusters, and a targeted control experiment can be performed to accelerate the structure-based drug development process.

[0158] Compared with the way of scoring by single docking and filtering by threshold, the embodiment based on molecular structure, skeleton, and interaction can be less affected by the accuracy of the docking scoring algorithm, and can comprehensively consider the structural information of the combination of the ligand molecule and the receptor molecule, and is more consistent with the idea of drug designers to promote drug development.

[0159] Compared with single-dimension docking scoring, the embodiment splits the molecular screening process into three dimensions of “skeleton clustering”, “structure clustering”, and “interaction clustering” in sequence, and the information amount is more.

[0160] In the process of multi-dimensional classification and aggregation, the embodiment takes “skeleton clustering” as the first, and the purpose is to let the whole screening process take the difference of molecular skeleton as the root category, which is more consistent with the development habit of drug designers. It should be noted that the process of classification and aggregation of the two dimensions of “structure clustering” and “interaction clustering” has no sequence requirement.

[0161] The embodiment verifies the stability of the interaction (binding mode) in a kinetic simulation manner, which guarantees the reliability of the introduction of the interaction fingerprint.

[0162] Another aspect of the present application also provides a design method.

[0163] Figure 10 A flowchart of a design method according to an embodiment of the present application is schematically shown.

[0164] As shown in Figure 10 the design method includes operation S1010 to operation S1020.

[0165] In operation S1010 , the molecular screening results are displayed, where the molecular screening results are screening results obtained according to the above method.

[0166] In operation S1020 , drug design or material design is performed based on the molecular screening results.

[0167] It should be noted that the process of screening molecules can refer to the relevant content shown above and will not be described in detail here.

[0168] Another aspect of the present application provides a device for screening molecules.

[0169] Figure 11 A block diagram of a device for screening molecules according to an embodiment of the present application is schematically shown.

[0170] like Figure 11 As shown, the device 1100 for screening molecules may include: a first mapping relationship obtaining module 1110 , a molecular skeleton extracting module 1120 , a molecular skeleton aggregating module 1130 and a second mapping relationship determining module 1140 .

[0171] The first mapping relationship obtaining module 1110 is used to obtain a first mapping relationship between the simplified molecular linear formulas of M ligand molecules and N molecular structures, where the simplified molecular linear formulas of the M ligand molecules each have structural information, and M and N are integers greater than or equal to 1.

[0172] The molecular skeleton extraction module 1120 is used to perform skeleton extraction on the structural information of at least part of the simplified molecular linear formulas of M ligand molecules, respectively, to obtain O molecular skeletons, where O is an integer greater than or equal to 1 and O is less than or equal to M.

[0173] The molecular skeleton aggregation module 1130 is used to aggregate O molecular skeletons to obtain P molecular skeleton types, where P is an integer greater than or equal to 1, and P is less than or equal to O.

[0174] The second mapping relationship determination module 1140 is used to determine a second mapping relationship between P molecular skeleton classes and N molecular structures based on the first mapping relationship, so as to screen ligand molecules that match the target receptor molecules based on the second mapping relationship.

[0175] In some embodiments, the apparatus 1100 further includes: a skeleton-based molecular structure acquisition module, a volume difference determination module, a structure clustering module, and a third mapping relationship determination module.

[0176] The skeleton molecular structure acquisition module is used to acquire the molecular structure corresponding to at least part of the ligand molecules corresponding to each of the P molecular skeleton classes, at least part of the skeleton classes.

[0177] The volume difference determination module is used to determine the volume difference between at least some of the ligand molecules based on the molecular structures corresponding to at least some of the ligand molecules.

[0178] The structural clustering module is used to cluster at least part of the ligand molecules corresponding to the skeleton class based on volume differences to obtain multiple structural classes.

[0179] The third mapping relationship determination module is used to determine a third mapping relationship between a plurality of structural classes and molecular structures based on the first mapping relationship.

[0180] In some embodiments, the volume difference determination module includes: a grid division unit, an occupancy ratio determination unit, an occupancy vector construction unit, and a volume difference determination unit.

[0181] The meshing unit is used to divide the pocket region of the target receptor molecule into a mesh.

[0182] The occupancy ratio determination unit is used to determine the occupancy ratio of the grid by the molecular structures corresponding to at least some of the ligand molecules.

[0183] The occupation vector construction unit is used to construct grid space occupation vectors of the molecular structures corresponding to at least some of the ligand molecules based on the occupation ratios.

[0184] The volume difference determining unit is configured to determine the volume differences between at least some of the ligand molecules based on grid space occupation vectors of molecular structures corresponding to at least some of the ligand molecules.

[0185] In some embodiments, the volume difference determining unit includes: a distance determining subunit and a volume difference determining subunit.

[0186] The distance determination subunit is used to determine the distance between the grid space occupation vectors of the molecular structures corresponding to two ligand molecules.

[0187] The volume difference determination subunit is used to determine the volume difference of the occupied space of the molecular structures corresponding to the two ligand molecules based on the distance between the grid space occupation vectors.

[0188] In some embodiments, the apparatus 1100 further includes: an interaction feature acquisition module, an interaction difference determination module, an interaction difference clustering module, and an action mapping relationship determination module.

[0189] The interaction feature acquisition module is used to obtain, for each class in at least some of the structural classes in multiple structural classes, a first interaction feature between each at least part of the molecular structure corresponding to the structural class and the target receptor molecule, and / or, for each class in at least some of the skeleton classes in P molecular skeleton classes, to obtain a second interaction feature between each at least part of the molecular structure corresponding to the skeleton class and the target receptor molecule.

[0190] The interaction difference determination module is used to determine first interaction differences between at least some of the first interaction features of the molecular structures and / or to determine second interaction differences between at least some of the second interaction features of the ligand molecules.

[0191] The interaction difference clustering module is used to cluster the first interaction differences to obtain a plurality of first interaction classes, and / or cluster the second interaction differences to obtain a plurality of second interaction classes.

[0192] The action mapping relationship determination module is configured to determine a fourth mapping relationship between multiple first interaction classes and molecular structures based at least on the first mapping relationship, and / or determine a fifth mapping relationship between multiple second interaction classes and molecular structures based on the first mapping relationship.

[0193] In some embodiments, the interaction difference determination module includes: an interaction feature vector determination unit and an interaction difference determination unit.

[0194] The interaction feature vector determining unit is configured to determine an interaction feature vector corresponding to the second interaction feature.

[0195] The interaction difference determination unit is used to repeat the following operations until the interaction difference between any two of the second interaction characteristics of at least some of the ligand molecules is determined: determining the distance between the interaction feature vectors corresponding to the second interaction characteristics; determining the second interaction difference of each of the two ligand molecules based on the distance between the interaction feature vectors corresponding to the second interaction characteristics.

[0196] In certain embodiments, the apparatus 1100 further includes: a representative molecule determination module, a stability feature acquisition module, and a stability mapping relationship determination module.

[0197] The representative molecule determination module is used to determine the representative molecule of the current class for any one of each first interaction class or each second interaction class.

[0198] The stability feature acquisition module is used to perform molecular dynamics simulation on the representative molecules of the current class to obtain the stability features of the representative molecules.

[0199] The stability mapping relationship determination module is configured to determine a sixth mapping relationship between the stability feature and the first interaction class or the second interaction class based on the stability feature of the representative molecule.

[0200] In some embodiments, the apparatus 1100 further includes an association storage module configured to store the simplified molecular linear formula, the molecular structure of the ligand molecule, and at least one of the following: the molecular skeleton class, the structure class, the first interaction class, or the second interaction class in association, to obtain a mapping table.

[0201] In some embodiments, the apparatus 1100 further includes a same-class merging module configured to perform same-class merging on the P molecular skeleton classes to obtain a multi-level molecular skeleton set, wherein a parent molecular skeleton in the multi-level molecular skeleton set corresponds to at least one child molecular skeleton, a bottom-level molecular skeleton corresponds to at least one ligand molecule, and the skeleton structure of the child molecular skeleton is more complex than the skeleton structure of the parent molecular skeleton.

[0202] In some embodiments, the apparatus 1100 further includes a skeleton graph generation module configured to generate a skeleton graph, wherein the skeleton graph includes a plurality of nodes, a non-terminal node in the plurality of nodes represents at least part of the molecular skeleton in the multi-level molecular skeleton set, a terminal node in the plurality of nodes represents a molecular cluster in the M ligand molecules that includes the skeleton corresponding to the terminal node, and one parent node in the plurality of nodes corresponds to at least one child node.

[0203] Another aspect of the present application also provides an apparatus for evaluating molecules.

[0204] Figure 12 An apparatus for evaluating molecules according to an embodiment of the present application is schematically shown in a block diagram.

[0205] Referring to Figure 12 The apparatus 1200 for evaluating molecules can include a simplified molecular linear formula obtaining module 1210, a to-be-screened molecular skeleton obtaining module 1220, and a molecular evaluation module 1230.

[0206] The simplified molecular linear formula obtaining module 1210 is configured to obtain a simplified molecular linear formula of a to-be-screened molecule.

[0207] The to-be-screened molecular skeleton obtaining module 1220 is configured to determine a skeleton of the to-be-screened molecule based on the simplified molecular linear formula of the to-be-screened molecule.

[0208] The molecular evaluation module 1230 is configured to evaluate the to-be-screened molecule based on the skeleton of the to-be-screened molecule and a plurality of mapping relationships determined according to the apparatus 1100, wherein the plurality of mapping relationships include at least one of the first mapping relationship to the sixth mapping relationship.

[0209] Another aspect of the present application provides a design device.

[0210] Figure 13 A block diagram of a design device according to an embodiment of the present application is shown schematically.

[0211] Referring to Figure 13 The design device 1300 can include a screening result display module 1310 and a design module 1320.

[0212] The screening result display module 1310 is configured to display a molecular screening result, which is a screening result obtained based on the device 1100 described above.

[0213] The design module 1320 is configured to perform drug design or material design based on the molecular screening result.

[0214] As to the device 1100 for screening molecules, the device 1200 for evaluating molecules, and the device 1300 for designing in the above-described embodiments, the specific manners in which the respective modules and units perform operations have been described in detail in the embodiments of the method, and thus will not be described in detail here.

[0215] Another aspect of the present application provides an electronic device.

[0216] Figure 14 A block diagram of an electronic device for implementing a method of screening molecules according to an embodiment of the present application is shown schematically.

[0217] Referring to Figure 14 The electronic device 1400 includes a memory 1410 and a processor 1420.

[0218] The processor 1420 can be a central processing unit (CPU), and can also be other general-purpose processors, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, or the like. The general-purpose processor can be a microprocessor, or the processor can also be any conventional processor.

[0219] Memory 1410 may include various types of storage units, such as system memory, read-only memory (ROM), and permanent storage. ROM may store static data or instructions required by processor 1420 or other modules of the computer. Permanent storage may be a readable and writable storage device. A permanent storage device may be a non-volatile storage device that retains stored instructions and data even when the computer is powered off. In some embodiments, the permanent storage device uses a large-capacity storage device (e.g., a magnetic or optical disk, flash memory) as the permanent storage device. In other embodiments, the permanent storage device may be a removable storage device (e.g., a floppy disk, optical drive). System memory may be a readable and writable storage device or a volatile readable and writable storage device, such as dynamic random access memory. System memory may store some or all instructions and data required by the processor during operation. In addition, memory 1410 may include any combination of computer-readable storage media, including various types of semiconductor memory chips (e.g., DRAM, SRAM, SDRAM, flash memory, programmable read-only memory), and magnetic disks and / or optical disks may also be used. In some embodiments, the memory 1410 may include a readable and / or writable removable storage device, such as a compact disc (CD), a read-only digital versatile disc (e.g., DVD-ROM, dual-layer DVD-ROM), a read-only Blu-ray disc, an ultra-density optical disc, a flash memory card (e.g., SD card, mini SD card, Micro-SD card, etc.), a magnetic floppy disk, etc. Computer-readable storage media do not include carrier waves and transient electronic signals transmitted wirelessly or wired.

[0220] The memory 1410 stores executable codes. When the executable codes are processed by the processor 1420 , the processor 1420 may execute part or all of the above-mentioned methods.

[0221] In addition, the method according to the present application may also be implemented as a computer program or a computer program product, which includes computer program code instructions for executing some or all of the steps in the above method of the present application.

[0222] Alternatively, the present application can also be implemented as a computer-readable storage medium (or non-transitory machine-readable storage medium or machine-readable storage medium) on which executable code (or computer program or computer instruction code) is stored. When the executable code (or computer program or computer instruction code) is executed by a processor of an electronic device (or server, etc.), the processor executes part or all of the steps of the above-mentioned method according to the present application.

[0223] Having described various embodiments of the application, it is to be understood that the above description is meant to be illustrative only, and that many modifications and variations of the embodiments described herein are possible. It is therefore to be understood that within the scope of the appended claims, and their equivalents, many alternatives to the embodiments described herein are possible. The selection of terms to be used in the description is not intended to limit the scope of the embodiments described herein, but rather to best explain the principles of the embodiments, practical application, or improvement over the technology in the art, or to enable others skilled in the art to understand the embodiments disclosed herein.

Claims

1. A method for screening molecules, characterized in that The method comprises: Obtaining a first mapping relationship between simplified molecular linear formulas of M ligand molecules and N molecular structures, wherein each simplified molecular linear formula of the M ligand molecules has structural information, and M and N are integers greater than or equal to 1; each molecular structure is a structure after successful docking with a target receptor molecule; For at least part of the simplified molecular linear formulas of the M ligand molecules, respectively perform skeleton extraction on the structural information of the simplified molecular linear formulas to obtain O molecular skeletons, where O is an integer greater than or equal to 1, and O is less than or equal to M; Polymerizing the O molecular skeletons to obtain P molecular skeleton species, where P is an integer greater than or equal to 1, and P is less than or equal to 0; Determining a second mapping relationship between the P molecular scaffold classes and the N molecular structures based on the first mapping relationship, so as to screen a ligand molecule that matches the target receptor molecule based on the second mapping relationship; After obtaining P molecular skeleton types, the method further includes: For each class of at least some of the skeleton classes in the P molecular skeleton classes, obtain the molecular structures corresponding to at least some of the ligand molecules corresponding to the skeleton class; determine the volume differences between the at least some of the ligand molecules based on the molecular structures corresponding to the at least some of the ligand molecules; cluster the at least some of the ligand molecules corresponding to the skeleton class based on the volume differences to obtain multiple structural classes; determine a third mapping relationship between the multiple structural classes and the molecular structures based on the first mapping relationship, so as to screen the ligand molecules that match the target receptor molecules based on the third mapping relationship; and / or For each class of at least some of the P molecular skeleton classes, obtain the second interaction characteristics between at least some of the ligand molecules corresponding to the skeleton class and the target receptor molecule; determine the second interaction differences between the second interaction characteristics of at least some of the ligand molecules; cluster the second interaction differences to obtain multiple second interaction classes; determine the fifth mapping relationship between the multiple second interaction classes and the molecular structure based on at least the first mapping relationship, so as to screen the ligand molecules that match the target receptor molecule based on the fifth mapping relationship.

2. The method according to claim 1, characterized in that The determining of the volume difference between two of the at least some of the ligand molecules based on the molecular structures corresponding to the at least some of the ligand molecules comprises: Divide the pocket region of the target receptor molecule into a grid; determining the proportion of the grid occupied by the molecular structures corresponding to at least some of the ligand molecules; constructing grid space occupation vectors of the molecular structures corresponding to at least some of the ligand molecules based on the occupation ratios; The volume difference between each of the at least some of the ligand molecules is determined based on the grid space occupation vectors of the molecular structures corresponding to each of the at least some of the ligand molecules.

3. The method according to claim 2, characterized in that The determining of the volume difference between two of the at least some of the ligand molecules based on the grid space occupation vectors of the molecular structures corresponding to the at least some of the ligand molecules comprises: Determine the distance between the grid space occupation vectors of the molecular structures corresponding to the two ligand molecules; Determining the volume difference of the space occupied by the molecular structures corresponding to the two ligand molecules based on the distance between the grid space occupation vectors; The above operation is repeated until the volume difference between the molecular structures corresponding to any two ligand molecules in the at least part of the ligand molecules is determined.

4. The method according to claim 1, wherein After obtaining the plurality of structural classes, the method further comprises: For each class in at least part of the plurality of structural classes, obtaining a first interaction feature between each of at least part of the molecular structures corresponding to the structural class and a target receptor molecule; determining first interaction differences between pairs of first interaction features of at least a portion of the molecular structures; clustering the first interaction differences to obtain a plurality of first interaction classes; A fourth mapping relationship between the plurality of first interaction classes and the molecular structure is determined based on at least the first mapping relationship.

5. The method according to claim 1 or 4, characterized in that Determining the second interaction difference between at least some of the respective second interaction characteristics of the ligand molecules comprises: determining an interaction feature vector corresponding to the second interaction feature; Repeat the following steps until the interaction difference between any two of the second interaction characteristics of at least some of the ligand molecules is determined: determining a distance between interaction feature vectors corresponding to the second interaction feature; A second interaction difference between each of the two ligand molecules is determined based on a distance between the interaction feature vectors corresponding to the second interaction feature.

6. The method according to claim 4, characterized in that After obtaining the plurality of first interaction classes, or after obtaining the plurality of second interaction classes, the method further includes: For each first interaction class or each second interaction class, determining a representative molecule of the current class; Performing molecular dynamics simulation on the representative molecule of the current class to obtain stability characteristics of the representative molecule; A sixth mapping relationship between the stability feature and the first interaction class or the second interaction class is determined based on the stability feature of the representative molecule.

7. The method according to claim 6, characterized in that Also includes: The simplified molecular linear formula of the ligand molecule, the molecular structure, and at least one of the following: a molecular skeleton class, a structural class, a first interaction class, or a second interaction class are stored in association to obtain a mapping table.

8. The method according to any one of claims 1 to 4, 6 to 7, characterized in that After obtaining P molecular skeleton types, the method further includes: The P molecular skeleton classes are merged into the same type to obtain a multi-level molecular skeleton set, wherein the parent molecular skeleton in the multi-level molecular skeleton set corresponds to at least one child molecular skeleton, the bottom molecular skeleton corresponds to at least one ligand molecule, and the skeleton structure of the child molecular skeleton is more complex than the skeleton structure of the parent molecular skeleton.

9. The method according to claim 8, characterized in that After obtaining the multi-level molecular skeleton set, the method further includes: Generate a skeleton graph, wherein the skeleton graph includes multiple nodes, non-terminal nodes among the multiple nodes represent at least part of the molecular skeleton in the multi-level molecular skeleton set, terminal nodes among the multiple nodes represent a molecular cluster of the M ligand molecules including the skeleton corresponding to the terminal node, and a parent node among the multiple nodes corresponds to at least one child node.

10. A method for evaluating a molecule, characterized in that include: Obtaining a simplified molecular linear formula of the molecule to be screened; Determining the skeleton of the molecule to be screened based on the simplified molecular linear formula of the molecule to be screened; The molecule to be screened is evaluated based on the skeleton of the molecule to be screened and multiple mapping relationships determined according to the method according to any one of claims 1 to 9, wherein the multiple mapping relationships include: at least one of the first mapping relationship to the sixth mapping relationship.

11. A design method, characterized in that: The method comprises: Displaying molecular screening results, wherein the molecular screening results are screening results obtained by the method according to any one of claims 1 to 10; Drug design or material design is performed based on the molecular screening results.

12. A device for screening molecules, characterized in that include: a first mapping relationship obtaining module, configured to obtain a first mapping relationship between simplified molecular linear formulas of M ligand molecules and N molecular structures, wherein the simplified molecular linear formulas of the M ligand molecules each have structural information, and M and N are integers greater than or equal to 1; and each of the molecular structures is a structure after successful docking with a target receptor molecule; a molecular skeleton extraction module, configured to perform skeleton extraction on structural information of at least a portion of the simplified molecular linear formulas of the M ligand molecules, respectively, to obtain O molecular skeletons, where O is an integer greater than or equal to 1, and O is less than or equal to M; A molecular skeleton polymerization module, configured to polymerize the O molecular skeletons to obtain P molecular skeleton types, where P is an integer greater than or equal to 1, and P is less than or equal to 0; a second mapping relationship determination module, configured to determine a second mapping relationship between the P molecular scaffold classes and the N molecular structures based on the first mapping relationship, so as to screen a ligand molecule that matches the target receptor molecule based on the second mapping relationship; The apparatus further includes a third mapping relationship determination module configured to obtain, for each of at least some of the P molecular skeleton classes, molecular structures corresponding to at least some of the ligand molecules corresponding to the skeleton class; and determine volume differences between pairs of the at least some of the ligand molecules based on the molecular structures corresponding to the at least some of the ligand molecules; Clustering at least some of the ligand molecules corresponding to the scaffold class based on the volume difference to obtain a plurality of structural classes; determining a third mapping relationship between the plurality of structural classes and the molecular structure based on the first mapping relationship, so as to screen a ligand molecule that matches the target receptor molecule based on the third mapping relationship; and / or The apparatus further includes an action mapping relationship determination module for obtaining, for each of at least some of the P molecular scaffold classes, second interaction characteristics between at least some of the ligand molecules corresponding to the scaffold class and the target receptor molecule; determining a second interaction difference between at least a portion of the respective second interaction characteristics of the ligand molecules; The second interaction differences are clustered to obtain a plurality of second interaction classes; and a fifth mapping relationship between the plurality of second interaction classes and the molecular structure is determined based at least on the first mapping relationship, so as to screen a ligand molecule that matches the target receptor molecule based on the fifth mapping relationship.

13. A device for evaluating molecules, characterized in that include: A simplified molecular linear formula obtaining module is used to obtain a simplified molecular linear formula of the molecule to be screened; A molecular skeleton obtaining module for screening, used for determining the skeleton of the molecule to be screened based on the simplified molecular linear formula of the molecule to be screened; A molecule evaluation module is used to evaluate the molecule to be screened based on the skeleton of the molecule to be screened and multiple mapping relationships determined by the device according to claim 12, wherein the multiple mapping relationships include: at least one of the first mapping relationship to the sixth mapping relationship.

14. A design device, characterized in that include: A screening result display module, for displaying molecular screening results, wherein the molecular screening results are screening results obtained by the device according to claim 12; A design module is used to perform drug design or material design based on the molecular screening results.

15. An electronic device, characterized in that: include: processor; as well as A memory having executable codes stored thereon, which, when executed by the processor, causes the processor to execute the method according to any one of claims 1 to 11.

16. A computer-readable storage medium, characterized in that An executable code is stored thereon, and when the executable code is executed by a processor of an electronic device, the processor is caused to execute the method according to any one of claims 1 to 11.

Images