Molecular design and structure optimization method, system and device and storage medium

Abstract

The invention provides a molecular design and structure optimization method, system and device and a storage medium. According to the method provided by the invention, a fragment-based drug design method is adopted, and automatic design, optimal screening and evaluation of virtual series molecules are carried out; for a molecular building block database, a series of molecules are designed and evaluated by adopting the algorithm, methods such as protein-small molecule docking, molecular drug properties, poor pharmacokinetic properties or improper toxic reaction (ADME / T) are adopted, and fragment composition of the series of molecules is continuously optimized by applying a related global optimization algorithm. A series of molecules which are considered to be good or optimal by an evaluation function can be generated and are used for guiding research and development of small-molecule drugs.

Description

Technical Field

[0001] The present invention belongs to the technical field of medicinal chemistry, and particularly relates to a molecular design and structure optimization method, system, device and storage medium. Background Art

[0002] Drug discovery is a complex research and exploration process integrating multiple disciplines. Its main steps can be briefly summarized as a cyclic and spiral upward pattern from genes to proteins, from proteins to lead compounds, and then optimizing and evaluating the lead compounds. In this process, computers have long become an indispensable means; for example, technologies such as molecular simulation, combinatorial chemistry, structure-activity relationship, and data mining are all advantages of computer-aided drug research and development. Using computers to search for and optimize lead compounds by various possible methods and improve their physicochemical indexes and biological activities has become one of the key steps in the drug discovery process.

[0003] Drug discovery usually faces problems of high cost and long cycle. A successful candidate drug must not only meet the biological activity criteria, but also have good pharmacokinetic properties, no obvious toxic and side effects, and good synthetic accessibility at the same time.

[0004] Series molecular design is a challenging problem in chemoinformatics. In the lead optimization stage, the above properties are improved by making structural modifications to the lead compound. Its basic logic is to analyze the structure-activity relationship (SAR). Without changing the molecular core of the lead compound, accurate SAR can be obtained by comprehensively comparing side chain substituents, and then these substituents are gradually adjusted to obtain a compound structure with better properties. If a structure database similar to the given molecular structure can be obtained quickly, the efficiency of lead optimization can be greatly improved. At present, there are already some methods to quickly generate a structure database similar to the given structure, such as generative models based on deep learning, analysis methods based on matched molecular pairs (MMPs), and some methods based on molecular graphs, etc.

[0005] Computer-aided drug design (CADD) is a powerful and promising technology that can be used for faster, cheaper, and more effective drug design. Using public molecular databases and CADD to screen for active compounds has become a common method. However, the compounds screened in public molecular databases often face difficulties in synthesis or are unavailable for purchase. Moreover, there are a large number of small molecular fragments (such as monocyclic aromatic hydrocarbons, polycyclic aromatic hydrocarbons, short-chain aliphatic hydrocarbons, etc.) in public databases, which are usually difficult to use. In addition, the structural diversity and quantity in public databases are also limited, making it difficult to cover a larger chemical space. Therefore, a better way needs to be found to build a larger synthetic chemical space and establish a set of methods for mining large-scale chemical spaces to solve the above problems.

[0006] The chemical space of small molecule compounds is extremely vast, with the number of synthesizable organic molecules alone exceeding 10 60 orders of magnitude. Although humans have only explored a tiny region of this vast space so far, this exploration has already produced a large number of drug molecules. Therefore, more effective chemical space exploration methods can help address the challenges of many diseases faced by humans. Traditional chemical space exploration methods aim to exhaustively list a series of compound molecules with well-defined structures. In recent years, deep generative models have become tools for chemical space exploration, using deep neural networks to learn the chemical properties implicit in a set of training set molecules and randomly sampling molecules with entirely new structures from the learned target chemical space. Currently, regarding the methods for constructing large-scale chemical spaces, the deep generative methods to address this challenge can be divided into two major categories according to the different representations of molecules. The first type of model uses string representations of molecules, such as the Simplified molecular input lineentry specification (SMILES), where atoms and chemical bonds are represented as characters. Through this representation, a character-based language model can be trained using a Recurrent Neural Network (RNN) architecture. The second type of model is directly based on the graph representation of molecules, using RNN sequential encoding or Graph Neural Network (GNN) to encode in a permutation-invariant manner.

[0007] However, current deep learning-based molecular generation methods can rapidly explore a vast chemical space, but at the same time, they also face some problems: such as low structural drug-likeness; poor synthetic accessibility; generating a large number of molecular structures (more than tens of billions), which incurs high storage and management costs. Graph-based models are more expressive in terms of rules. However, graph-based models require a large amount of computational resources for training and are less efficient. Moreover, due to the sequential nature of SMILES strings, it is difficult to generate molecules that retain the structural core. A large number of chemically invalid and repetitive molecules are often generated during the generation process. It is difficult to analyze the structure-activity relationship, and it is even more difficult to rapidly prepare molecules through automated batch synthesis methods.

[0008] Combinatorial chemistry (combichem) synthesizes a large number of molecularly diverse compounds in parallel and systematically through the combination and connection of molecular building blocks to form a combinatorial library. Its goal is to synthesize a large number of various compounds like building blocks to meet the demand for diverse compounds in modern drug research.

[0009] In the early stage of the development of combinatorial chemistry, the construction of combinatorial libraries was mostly random, that is, randomly selecting molecular building blocks and obtaining a library containing a large number of compounds after a series of syntheses. Since this method of constructing libraries has no purpose, it is a great waste of manpower, material resources, and time.

[0010] Combinatorial chemistry provides a method for rapidly establishing a large-scale compound molecular library. However, if this task can be virtually completed by a computer for further screening, optimization, etc., it will be more flexible and resource-saving. It should be emphasized that the compound molecular library thus combined is not composed of random molecular structures but requires careful design.

[0011] The purpose of evaluating the molecular diversity of combinatorial libraries is to reduce costs and increase the probability of discovering new active compounds. It is hoped to use computer tools to select compounds for purchase or synthesis, improve the molecular diversity of compound libraries in activity screening, and also hope to design diverse compound libraries targeting the three-dimensional structure of specific targets through computer tools, thereby obtaining compounds that meet safety, efficiency, economy, and activity requirements.

[0012] The close combination of virtual combinatorial chemistry and computational chemistry has given combinatorial chemistry stronger vitality. One trend in the development of combinatorial chemistry is to combine it with rational drug design. Before synthesizing a combinatorial library, a virtual compound library is rationally designed through molecular simulation and theoretical calculation methods to increase the diversity and drug-likeness of the compounds in the library and improve the quality of the library. Many commercial software contains some combinatorial chemistry library modules. For example, the Colibri software (BioSolveIT Ltd.) can be used for the design of combinatorial chemistry libraries, and the Sybyl software also integrates a virtual combinatorial library generation module, but neither of them has a complete library evaluation function.

[0013] Virtual combinatorial chemistry databases often build databases with existing molecular building blocks as the core. After the virtual combinatorial chemistry database is established, virtual screening methods are used to predict how compounds interact with a given target protein. However, this method can only use CADD methods for screening after enumerating the combinatorial chemistry library to generate a sub-library of complete molecules. It cannot be widely used, lacks flexibility, and does not have an automated feedback optimization function.

[0014] Sadybekov and his colleagues developed a method called V-SYNTHES (virtual synthon hierarchical enumeration screening), which reduces the number of molecules that need to be evaluated when searching for potential hits in these libraries, and the computational resources used are 1 / 100 of the standard method.

[0015] The FlexX chemical space docking algorithm of GeneTech Company is similar to the V-SYNTHES method. It is also a docking algorithm based on fragment increment. Adopting the idea of FBDD, it starts from fragments, first docks the fragments, and extends on the basis of the fragments.

[0016] Using the above methods can adapt to large-scale virtual combinatorial chemistry libraries and can perform molecular docking mining on large-scale libraries. However, the disadvantages of the above methods are also very prominent. First of all, the above methods can mine chemical spaces at the tens of billions level, but may not be able to mine larger-scale libraries; finding anchor fragments may not be able to truly reflect the performance of the fragments in the complete molecule according to the docking results; the method for evaluating the constituent fragments depends on the docking results, and the evaluation method is relatively single and cannot handle multiple scenarios; the scoring function setting is not flexible, and it is difficult to achieve selective molecule and drug-like molecule design; evaluating the constituent fragments depends on the optimal docking score, and the score of the complete molecule after enumeration is not necessarily the best.

[0017] Therefore, it is necessary to establish a chemical space and an automated feedback optimization mining method that can utilize existing small molecule fragments, is easy to synthesize, has a wide range of applications, and enables researchers to discover new active small molecules through CADD. Summary of the Invention

[0018] Aiming at the deficiencies of the prior art, the purpose of the present invention is to provide a molecular design and structure optimization method, system, device and storage medium. Its main purpose is to adopt a fragment-based drug design method based on given evaluation indicators to automatically design, optimize, screen and evaluate virtual series molecules; for the molecular building block database, use this algorithm to design and evaluate series molecules, and use methods such as protein-small molecule docking, molecular drug-likeness, adverse pharmacokinetic properties or inappropriate toxicity reactions (ADME / T), etc., and apply relevant global optimization algorithms to continuously optimize the fragment composition of the series molecules, which can generate series molecules considered better or optimal by the evaluation function for guiding the research and development of small molecule drugs.

[0019] To achieve the purpose of this invention, the present invention adopts the following technical solutions:

[0020] In the first aspect, the present invention provides a molecular design and structure optimization method, and the method includes:

[0021] Obtain molecular building block data, standardize the SMILES structure of the molecular building blocks, and establish a molecular building block database;

[0022] Based on prior knowledge, establish chemical reaction rules, write a chemical reaction rule module, and write a reaction classification module;

[0023] According to the chemical reaction rule module and the reaction classification module, split the reaction sites of the input molecular building block structure to obtain decomposition fragments, and perform reaction classification marking; the decomposition fragments retain the molecular parent nucleus structure information, and calculate and mark the molecular parent nucleus attribute information;

[0024] Based on the molecular parent nucleus attribute information, through the data of the molecular building block database, sort and number the molecular building block fragments by attribute values, and select several molecular building block fragments that make up a series of molecules;

[0025] Based on the several molecular building block fragments that make up a series of molecules and the chemical reaction rules, perform structure recombination through chemical reaction product enumeration to obtain several target series molecule structures;

[0026] Perform 3D structure preparation on the several target molecule structures to obtain corresponding 3D conformations;

[0027] Perform protein-small molecule docking and / or pharmacodynamic parameter evaluation processing on the 3D conformation and the target protein, and preliminarily select potential active molecules or druggable molecules;

[0028] Through a global optimization algorithm, iteratively update the selected potential active molecules or druggable molecules, and output a better target series of molecules.

[0029] Preferably, the obtaining of molecular building block data, the normalization processing of the SMILES structure of molecular building blocks, and the establishment of a molecular building block database include:

[0030] Perform structure preparation on all molecules in the molecular building block data and save them in a preset format.

[0031] Preferably, based on prior knowledge, establish chemical reaction rules, write a chemical reaction rule module, and write a reaction classification module:

[0032] Establish chemical reaction rules for drug synthesis based on organic chemistry textbooks and literature;

[0033] The chemical reaction rules are represented by SMARTS expressions.

[0034] Preferably, the molecular parent nucleus attribute information includes molecular weight, number of rotatable bonds, hydrogen bond acceptors, hydrogen bond donors, topological polar surface area, octanol / water partition coefficient, QED, and MCE-18.

[0035] Preferably, the sorting number is selected by the global optimization algorithm.

[0036] Preferably, the enumeration performs structural reorganization by means of fragment connection, while marking the reaction type and whether it can react.

[0037] Preferably, the protein-small molecule docking and pharmacodynamic parameter evaluation processing can be carried out using conventional molecular docking scoring software (such as common CADD software like Vina, CB-Dock, and Schrödinger, etc.)

[0038] Preferably, the pharmacodynamic parameters include molecular weight, number of rotatable bonds, hydrogen bond acceptors, hydrogen bond donors, topological polar surface area, octanol / water partition coefficient, QED, and ADME / T prediction models.

[0039] Preferably, the global optimization algorithm includes genetic algorithm, simulated annealing algorithm, tabu search algorithm, particle swarm algorithm, or ant colony algorithm.

[0040] Preferably, after outputting the better target series of molecules, it further includes generating a report of the better target series of molecules, and the report includes molecular structure, composition fragment information, molecular building block information, physicochemical properties, and Docking scores.

[0041] In a second aspect, the present invention provides a molecular design and structure optimization system, comprising:

[0042] A first module: obtaining molecular building block data, standardizing the SMILES structure of molecular building blocks, and establishing a molecular building block database;

[0043] A second module: establishing chemical reaction rules based on prior knowledge, writing a chemical reaction rule module, and writing a reaction classification module;

[0044] A third module: splitting the reaction sites of the input molecular building block structure according to the chemical reaction rule module and the reaction classification module to obtain decomposition fragments, and performing reaction classification marking; the decomposition fragments retain the molecular parent nucleus structure information, and calculate and mark the molecular parent nucleus attribute information;

[0045] A fourth module: based on the molecular parent nucleus attribute information, sorting and numbering the molecular building block fragments through the data in the molecular building block database, and selecting several molecular building block fragments that make up a series of molecules;

[0046] A fifth module: based on the molecular building block fragments of the several series of molecules and the chemical reaction rules, performing structure recombination through chemical reaction product enumeration to obtain several target series of molecular structures;

[0047] A sixth module: preparing the 3D structure of the several target molecular structures to obtain corresponding 3D conformations;

[0048] A seventh module: performing protein-small molecule docking and / or pharmacodynamic parameter evaluation processing on the 3D conformation and the target protein, and preliminarily selecting potential active molecules or druggable molecules;

[0049] An eighth module: iteratively updating the selected potential active molecules and / or druggable molecules through a global optimization algorithm, and outputting relatively optimal target series of molecules;

[0050] A ninth module: generating a report of relatively optimal target series of molecules.

[0051] Preferably, in the first module, the SMILES structure of the molecular building block is standardized, and the processing software is Openbabel, the RDKit package of the Python language, etc.

[0052] Preferably, in the second module, the intelligent reaction classification module is written based on the intelligent reaction module in the RDKit program in Python.

[0053] Preferably, in the third module, the reaction sites of the input molecular building block structure are split, and it is written based on the RDKit program package in Python.

[0054] Preferably, in the third module, the RDKit program in Python is used to calculate the property information of molecular building blocks.

[0055] Preferably, in the fifth module, the chemical reaction product enumeration step is written based on the RDKit program package. This enumeration process is carried out by means of fragment connection, while marking the reaction type and whether the reaction can occur.

[0056] Preferably, the generation of 3D conformations in the sixth module can be based on software such as the RDKit 3D structure generation module, GypSum or GeoMol in the Python program package.

[0057] Preferably, for the seventh module, the software for protein-small molecule docking and pharmacodynamic parameter evaluation is common CADD software such as Vina, CB-Dock and Schrödinger; the pharmacodynamic parameters include molecular weight, number of rotatable bonds, hydrogen bond acceptors, hydrogen bond donors, topological polar surface area, octanol / water partition coefficient, QED, ADME / T prediction models, etc.

[0058] Preferably, the global optimization algorithm in the eighth module is written based on packages such as HyperOpt and Tune in Python. The global optimization algorithm can select genetic algorithm, simulated annealing algorithm, tabu search algorithm, particle swarm algorithm or ant colony algorithm.

[0059] Preferably, in the ninth module, the generation of reports is based on packages such as Numpy, Pandas and RDKit in Python.

[0060] In a third aspect, the present invention provides the application of the molecular design and structure optimization method and system as described above in screening active small molecules, including the following:

[0061] (1) Construct the above-mentioned molecular building block database or the above-mentioned molecular design and structure optimization method and system;

[0062] (2) Select the required protein target, use docking software to score the molecules in the selected range, and through the molecular building block database, use the above-mentioned molecular design and structure optimization method and system to design several better molecular series;

[0063] (3) Through molecular dynamics simulation software, simulate several better molecular series designed, and screen out molecules with better simulation results;

[0064] (4) Through in vitro activity experiment evaluation, screen out small molecule compounds with better activity.

[0065] Preferably, the molecular docking software in the step (3) includes common molecular dynamics simulation software such as Gromacs and Amber;

[0066] Preferably, the in vitro activity experiment in the step (4) includes, but is not limited to, common in vitro activity experiments such as ADP-Glo and MTT.

[0067] In a fourth aspect, the present invention provides a molecular design and structure optimization device, including a processor and a memory; the memory is used to store programs; the processor executes the programs to implement the method as described above.

[0068] In a fifth aspect, the present invention further provides a computer-readable storage medium, the storage medium stores a program, and the program is executed by a processor to implement the method as described above.

[0069] Compared with the prior art, the present invention has the following beneficial effects:

[0070] As a basic tool for exploring the chemical space of molecular building blocks, the molecular design and structure optimization method provided by the present invention has broad application prospects. Drug discovery is essentially a cycle of screening-optimization-evaluation, and then screening-optimization-evaluation; in the current field of drug research and development, every step and every link is inseparable from the computer. The molecular design and structure optimization method provided by the present invention just provides such a function, which can simply and quickly let as many molecular structure series of interest enter the stage to be screened, and improve the success rate of screening.

[0071] The chemical space of the molecular building blocks of the present invention is a virtual combinatorial chemistry library made and stored by using computer-related software or programs. The chemical space is made by known chemical reactions and a molecular building block library through relevant software or algorithms. An ideal chemical space should include compounds of various structural types and molecular characteristics, that is, the molecules should have as large a difference as possible. For example, in drug design, there are mainly two problems in the design and application of chemical space. One is to determine which chemical structures are valuable in a specific chemical space; the other is to determine a series of molecules that can efficiently search and evaluate these chemical spaces.

[0072] Compared with the simple combinatorial chemistry experiment optimization in the prior art, the method for designing and optimizing the molecular building block database of the present invention has the advantages of high speed, and tens of thousands of compounds can be optimized in one day; low cost, without consuming any experimental equipment, and completely realized through computer simulation; strong information analysis means, and the automatic global optimization algorithm can provide the best series of molecular design results statistically. Therefore, the success rate of the virtual series of molecules designed by using the molecular design and structure optimization method provided by the present invention is high, and the probability that the better molecules obtained by secondary screening from the series of molecules have good activity is high.

[0073] The method of the present invention for a series of molecular design and optimization by combining a global optimization algorithm with a combinatorial chemistry method can be used for the research on the design of novel molecular structures based on the three-dimensional structure of protein targets, providing a new method for the application of combinatorial chemistry and computer-aided drug design in drug research.

[0074] The computer program of the molecular design and structure optimization method written by the present invention can, based on the given three-dimensional structure information of the receptor macromolecule, adopt a structure-based drug design method to automatically design virtual series molecules for a large-scale chemical space, and then evaluate the series molecules and the target protein by methods such as protein-small molecule docking, molecular drug-likeness, or ADME / T, etc., and apply a global optimization algorithm to continuously optimize the series molecules to optimize and generate series molecules with high affinity for the target, good drug-likeness, or other expectations.

[0075] The computer program of the molecular design and structure optimization method written by the present invention can, according to requirements, set the selection of the evaluation function by itself. After obtaining relevant series molecules, other property filtering or evaluation can be carried out. The present invention first constructs a molecular building block database with a huge quantity by using the method of virtual combinatorial chemistry; the database uses public or internal molecular building blocks as the basis and common chemical reactions as the conditions, and performs synthesis and splicing during the global optimization design process, which not only has a large order of magnitude, but also has good synthesizability, and thus has the characteristics of high cost performance. Therefore, the present invention is more realistic and has good application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0076] Figure 1 is the flow chart of the global optimization algorithm molecular design for establishing the molecular building block chemical space and the molecular design and structure optimization method provided by the embodiment;

[0077] Figure 2 is the main process flow chart for establishing the molecular building block chemical space;

[0078] Figure 3 is the step flow chart for molecular design using the molecular design and structure optimization method provided by the present invention;

[0079] Figure 4 is the histogram of the affinity frequency distribution of the molecular design in Example 1;

[0080] Figure 5 is the docking binding mode diagram for determining the parent nucleus molecule in Example 1;

[0081] Figure 6 is the docking three-dimensional conformation diagram of the molecular design and structure optimization screened in Example 1. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0082] The technical solution of the present invention will be further described below through specific embodiments. Those skilled in the art should understand that the embodiments are only helpful for understanding the present invention and should not be regarded as specific limitations on the present invention.

[0083] The molecular design and structure optimization method provided in the following embodiments, the flowchart refers to Figure 1 and the main process of establishing the chemical space of molecular building blocks refers to Figure 2 and the steps of molecular design using the molecular design and structure optimization method provided by the present invention refer to Figure 3 .

[0084] Specifically as follows:

[0085] S100. Obtain molecular building block data, standardize the SMILES structure of molecular building blocks, and establish a molecular building block database;

[0086] It should be noted that in some embodiments, based on a public structure database, collect molecular building block structure data, convert it into a standardized SMILES sequence through a biochemical toolkit; perform structure preparation on all molecules in the molecular building block structure data and save it in a preset format; the software for processing the molecular structural formula and SMILES format of molecular building block data is Openbabel and the RDKit package in the Python language;

[0087] Specifically, obtain high-quality molecular building block (BBs) structures from the database. The detailed steps are as follows: collect molecular building block structure data from each public structure database and convert it into a standardized SMILES sequence through RDKit (RDKit is a commonly used biochemical information Python toolkit. It provides a large number of computational operations on the 2D or 3D of chemical molecules and can generate molecular descriptors for machine learning). Among them, the public data blocks include databases such as Reaxys, PubChem, ChEMBL, and ZINC.

[0088] S200. Based on prior knowledge, establish chemical reaction rules, write a chemical reaction rule module, and write a reaction classification module;

[0089] It should be noted that in some embodiments, chemical reaction rules for drug synthesis are established based on organic chemistry textbooks and literature; among them, the chemical reaction rules are represented by SMARTS expressions; the reaction classification module is written based on the intelligent reaction module in the RDKit program in Python, and the molecular building blocks are classified by chemical reaction rules.

[0090] Specifically, to make the generated molecules have good synthesizability, common chemical reaction rules for drug synthesis are established from organic chemistry textbooks and literature and represented by SMARTS expressions. Among them, the chemical reaction rules take into account common types of drug synthesis reactions, including coupling reactions, esterification reactions, cycloaddition reactions, and ring formation reactions, etc. Each reaction rule is encoded with SMARTS, enabling the decomposition and recombination of molecular structures according to the reaction rules.

[0091] S300. According to the chemical reaction rule module and the reaction classification module, the reaction sites of the input molecular building block structure are segmented to obtain decomposition fragments, and reaction classification labels are given; the decomposition fragments retain the molecular parent nucleus structure information, and the molecular parent nucleus attribute information is calculated and labeled.

[0092] It should be noted that the segmentation of the reaction sites of the input molecular building block structure is written based on the RDKit program package in Python; the compound attributes include molecular weight, number of rotatable bonds, hydrogen bond acceptors, hydrogen bond donors, topological polar surface area, octanol / water partition coefficient, QED, and MCE-18, etc.; the attribute information of the molecular building blocks is calculated using the RDKit program package in Python.

[0093] S400. Based on the molecular parent nucleus attribute information, through the data in the molecular building block database, the molecular building block fragments are sorted and numbered by attribute values, and several molecular building block fragments for forming a series of molecules are selected.

[0094] Specifically, several molecular building block fragments for forming a series of molecules are selected according to the sorting number, and the selection of the sorting number is automatically selected by a global optimization algorithm.

[0095] S500. Based on the several molecular building block fragments for forming a series of molecules and the chemical reaction rules, structural recombination is carried out through chemical reaction product enumeration to obtain several target series of molecular structures.

[0096] It should be noted that the chemical reaction product enumeration step is written based on the RDKit program package, and this enumeration process is carried out by means of fragment connection, while marking the reaction type and whether the reaction can occur.

[0097] S600. 3D structure preparation is carried out on several target series of molecules to obtain corresponding 3D conformations.

[0098] It should be noted that the generation of small molecule 3D conformations can be based on software such as the 3D conformation generation module of the Python package RDKit, GypSum, or GeoMol;

[0099] Specifically, a large number of molecules similar to the given compound structure obtained in the specific embodiment of S500 are subjected to structure preparation through an internal molecular preparation process, including hydrogenation, protonation, etc., to generate three-dimensional conformations.

[0100] S700. For the 3D conformation, perform protein-small molecule docking or pharmacodynamic parameter evaluation processing with the target protein, and initially select potential active molecules or druggable molecules;

[0101] It should be noted that the molecular docking scoring software is common CADD software such as Vina, CB-Dock, and Schrödinger; the pharmacodynamic parameter evaluation includes molecular weight, number of rotatable bonds, hydrogen bond acceptors, hydrogen bond donors, topological polar surface area, octanol / water partition coefficient, QED, ADME / T prediction models, etc.

[0102] S800. Through a global optimization algorithm, iteratively update the selected molecular series, and execute modules 4-7 in the iterative order according to the set number of steps and conditions;

[0103] It should be noted that the global optimization algorithm module is written based on packages such as HyperOpt and Tune in Python, and the global optimization algorithm can select genetic algorithm, simulated annealing algorithm, tabu search algorithm, particle swarm algorithm, or ant colony algorithm, etc.

[0104] S900. After the optimization design is completed, generate a report file of the molecular series with better feedback for the user;

[0105] It should be noted that the result report includes contents such as molecular structure, composition fragment information, molecular building block information, physicochemical properties, and Docking scores. The generation of the result report is based on packages such as Numpy, Pandas, and RDKit in Python;

[0106] Specifically, after all tasks are completed, call data analysis libraries such as RDKit and Numpy, Pandas, and RDKit, directly process contents such as molecular structure, composition fragment information, molecular building block information, physicochemical properties, and docking scores in Python, and output the data analysis result report.

[0107] The following embodiments adopt the above process:

[0108] Embodiment 1

[0109] Cyclin-dependent kinase 2 (CDK2) is one of the main members of the cyclin-dependent kinase family of serine / threonine (Ser / Thr) protein kinases. Abnormal expression or function of CDK2 is related to the pathogenesis of various diseases such as tumors, viral replication and infection, immunodeficiency diseases, and male infertility. CDK2 inhibitors have become an important target in the research and development of anti-tumor drugs. To obtain CDK2 inhibitors with excellent properties, download the three-dimensional crystal structure of this target, and use the above-mentioned molecular design and structure optimization methods and molecular building block chemical space to optimize the design of potential active compounds. The design process includes the following steps:

[0110] First, for the data in the molecular building block database, sort and number the molecular building block fragments by attribute values, and select several molecular building block fragments that make up a series of molecules;

[0111] Second, based on the molecular building block fragments of the several series of molecules and chemical reaction rules, perform structure recombination through chemical reaction product enumeration to obtain several target series of molecular structures;

[0112] Third, perform structure preparation on the several target molecular structures to obtain the corresponding 3D conformations;

[0113] Fourth, dock the 3D conformations of the several molecular structures with the CDK2 target crystal structure (pdb format) for protein-small molecule docking, and initially select potential active series of molecules;

[0114] Fifth, through a global optimization algorithm, iteratively update the selected molecular series, and execute the above steps 1-4 in iteration order according to the set number of steps and conditions;

[0115] Sixth, after the optimization design is completed, generate a report file of the series of molecules with better feedback for experts to manually select and further design.

[0116] The optimized designed molecules are docked with the CDK2 target molecule to obtain the three-dimensional conformation and scoring ranking of the molecules, as well as the docking scoring frequency distribution histogram, as Figure 4 shown; determine the docking binding mode diagram of the parent nucleus molecule (taking molecule 3 as an example), as Figure 5 shown. Retain the molecules with a docking score less than -9 for experts to select the parent nucleus and improve the design. The chemical molecular structures of the retained part are shown below, and their three-dimensional conformations are as Figure 6 shown.

[0117]

[0118]

[0119] After expert selection of the parent nucleus and improvement of the design, several candidate compounds obtained were subjected to an in vitro CDK2 activity inhibition experiment.

[0120] The CDK2 kinase activity assay was performed using recombinant CDK / cyclin (CDK2 / cyclin E) purchased from Carna Biosciences. All activity assays were repeated in white low-volume 384-well plates. The reaction was carried out in 40 mM Tris-HCl at pH 7.5, containing 20 mM MgCl2, 0.1% BSA, and 50 μM DTT, and the reaction was initiated by adding ATP and substrate peptide. The 384-well plate was incubated at 37 °C for 30 minutes, followed by a two-step ADP-Glo assay. Finally, the plate was read using a Tecan Spark plate reader. The data was analyzed using curve fitting software (GraphPad Prism version 9). The IC50... was determined from a 12-point concentration curve using an ATP concentration approximately equal to Km.

[0121] The Ki value was calculated according to the following formula: Ki = IC50 / (1 + S / Km), where S is the measured ATP concentration and Km is the Michaelis-Menten constant of ATP.

[0122] The results of the relevant IC50 and Ki values of some compounds (compounds 1-10) screened in the present invention are shown in Table 1.

[0123] Table 1

[0124]

[0125] Among them, compounds 1, 3, 6, 8, and 10 showed excellent effects and can be used for further drug development.

[0126] The content of the method embodiments of the present invention is applicable to the relevant electronic device embodiments. The functions specifically implemented by the electronic device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those of the above methods.

[0127] Another aspect of the embodiments of the present invention also provides a computer-readable storage medium. The storage medium stores a program, and the program is executed by a processor to implement the method as described above.

[0128] The content of the method embodiments of the present invention is applicable to the embodiments of this computer-readable storage medium. The functions specifically implemented by the embodiments of this computer-readable storage medium are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those of the above methods.

[0129] An embodiment of the present invention also discloses a computer program product or a computer program. The computer program product or the computer program includes computer instructions, and the computer instructions are stored in a computer-readable storage medium. The processor of the computer device can read the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the foregoing method.

[0130] In some alternative embodiments, the functions / operations mentioned in the block diagrams may not occur in the order mentioned in the operation diagrams. For example, depending on the functions / operations involved, two consecutive blocks shown may actually be executed substantially simultaneously, or the blocks can sometimes be executed in the reverse order. In addition, the embodiments presented and described in the flowcharts of the present invention are provided by way of example for the purpose of providing a more comprehensive understanding of the technology. The disclosed methods are not limited to the operations and logical flows presented herein. Alternative embodiments are contemplated, in which the order of various operations is changed and the sub-operations described as part of a larger operation are executed independently.

[0131] In addition, although the present invention has been described in the context of functional modules, it should be understood that, unless otherwise stated to the contrary, one or more of the functions and / or features described may be integrated in a single physical device and / or software module, or one or more functions and / or features may be implemented in separate physical devices or software modules. It should also be understood that a detailed discussion of the actual implementation of each module is not necessary for understanding the present invention. Rather, considering the attributes, functions, and internal relationships of the various functional modules in the devices disclosed herein, the actual implementation of the modules will be understood within the ordinary skills of an engineer. Therefore, those skilled in the art can implement the present invention as set forth in the claims without undue experimentation. It should also be understood that the specific concepts disclosed are illustrative only and are not intended to limit the scope of the present invention, which is determined by the full scope of the appended claims and their equivalents.

[0132] If the described function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention.

[0133] The aforementioned storage medium includes: various media that can store program codes, such as USB flash drives, external hard drives, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical discs.

[0134] The logic and / or steps represented in the flowchart or described in other ways herein, for example, can be considered as a definite sequence list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by an instruction execution device, apparatus, or equipment (such as a computer-based device, a device including a processor, or other devices that can fetch and execute instructions from the instruction execution device, apparatus, or equipment), or in combination with these instruction execution devices, apparatus, or equipment. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transmit a program for use by or in combination with an instruction execution device, apparatus, or equipment.

[0135] More specific examples (non-exhaustive list) of computer-readable media include the following: an electrical connection part (electronic device) having one or more wirings, a portable computer disk cartridge (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). Additionally, a computer-readable medium can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpretation, or other suitable processing as necessary, and then stored in a computer memory.

[0136] It should be understood that various parts of the present invention can be implemented using hardware, software, firmware, or a combination thereof. In the above-described embodiments, multiple steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution device. For example, if implemented using hardware, as in another embodiment, any one or a combination of the following techniques well known in the art can be used: discrete logic circuits having logic gate circuits for implementing logical functions on data signals, application-specific integrated circuits having suitable combinational logic gate circuits, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0137] The applicant declares that the present invention illustrates the molecular design and structure optimization method, system, device and storage medium of the present invention through the above embodiments, but the present invention is not limited to the above embodiments, that is, it does not mean that the present invention must rely on the above embodiments to be implemented. Those skilled in the art should understand that any improvement of the present invention, the equivalent substitution of each raw material of the product of the present invention, the addition of auxiliary components, the selection of specific methods, etc. all fall within the protection scope and the disclosure scope of the present invention.

[0138] The preferred embodiments of the present invention have been described in detail above. However, the present invention is not limited to the specific details in the above embodiments. Within the technical concept scope of the present invention, various simple modifications can be made to the technical solution of the present invention, and these simple modifications all belong to the protection scope of the present invention.

[0139] In addition, it should be noted that, in the case of no contradiction, the various specific technical features described in the above specific embodiments can be combined in any suitable way. To avoid unnecessary repetition, the present invention will not separately describe various possible combination methods.

Claims

1. A method for molecular design and structure optimization, characterized in that, the method comprises: Obtaining molecular building block data, standardizing the SMILES structure of molecular building blocks, and establishing a molecular building block database; Based on prior knowledge, establishing chemical reaction rules, writing a chemical reaction rule module, and writing a reaction classification module; According to the chemical reaction rule module and the reaction classification module, splitting the reaction sites of the input molecular building block structure to obtain decomposition fragments, and performing reaction classification marking; the decomposition fragments retain the molecular parent nucleus structure information, and calculate and mark the molecular parent nucleus attribute information; Based on the molecular parent nucleus attribute information, sorting and numbering the molecular building block fragments by attribute values, and selecting several molecular building block fragments that make up a series of molecules; Based on the several molecular building block fragments that make up a series of molecules and the chemical reaction rules, performing structure recombination through chemical reaction product enumeration to obtain several target series of molecular structures; Performing 3D structure preparation on the several target molecular structures to obtain corresponding 3D conformations; Performing protein-small molecule docking and / or pharmacodynamic parameter evaluation processing on the 3D conformations and target proteins, and preliminarily selecting potential active molecules or druggable molecules; Through a global optimization algorithm, iteratively updating the selected potential active molecules or druggable molecules, and outputting better target series of molecules.

2. The method for molecular design and structure optimization according to claim 1, characterized in that, the obtaining molecular building block data, standardizing the SMILES structure of molecular building blocks, and establishing a molecular building block database includes: Performing structure preparation on all molecules in the molecular building block data and saving them in a preset format.

3. The method for molecular design and structure optimization according to claim 1 or 2, characterized in that, the based on prior knowledge, establishing chemical reaction rules, writing a chemical reaction rule module, and writing a reaction classification module: Establishing chemical reaction rules for drug synthesis based on organic chemistry textbooks and literature; The chemical reaction rules are represented by SMARTS expressions.

4. The method for molecular design and structure optimization according to any one of claims 1-3, characterized in that, the molecular parent nucleus attribute information includes molecular weight, number of rotatable bonds, hydrogen bond acceptors, hydrogen bond donors, topological polar surface area, octanol / water partition coefficient, QED, and MCE-18; Preferably, the sorting and numbering are selected by the global optimization algorithm.

5. The method for molecular design and structure optimization according to any one of claims 1-4, characterized in that, the enumeration for structure recombination is carried out by means of fragment connection, while marking the reaction type and whether it can react.

6. The method for molecular design and structure optimization according to any one of claims 1-5, characterized in that, the pharmacodynamic parameters include molecular weight, number of rotatable bonds, hydrogen bond acceptors, hydrogen bond donors, topological polar surface area, octanol / water partition coefficient, QED, and ADME / T prediction model.

7. The method for molecular design and structure optimization according to any one of claims 1-6, characterized in that, The global optimization algorithm includes genetic algorithm, simulated annealing algorithm, tabu search algorithm, particle swarm optimization algorithm or ant colony algorithm; Preferably, after outputting the better target series of molecules, a report of the better target series of molecules is further generated, and the report includes molecular structure, composition fragment information, molecular building block information, physicochemical properties and Docking scores.

8. A molecular design and structure optimization system Characterized in that It includes: The first module: obtaining molecular building block data, standardizing the SMILES structure of the molecular building blocks, and establishing a molecular building block database; The second module: based on prior knowledge, establishing chemical reaction rules, writing a chemical reaction rule module, and writing a reaction classification module; The third module: according to the chemical reaction rule module and the reaction classification module, splitting the reaction sites of the input molecular building block structure to obtain decomposition fragments, and performing reaction classification marking; the decomposition fragments retain the molecular parent nucleus structure information, and calculate and mark the molecular parent nucleus attribute information; The fourth module: based on the molecular parent nucleus attribute information, sorting and numbering the molecular building block fragments by attribute values, and selecting several molecular building block fragments that make up a series of molecules; The fifth module: based on the molecular building block fragments of the several series of molecules and the chemical reaction rules, performing structure recombination through chemical reaction product enumeration to obtain several target series of molecular structures; The sixth module: performing 3D structure preparation on the several target molecular structures to obtain corresponding 3D conformations; The seventh module: performing protein-small molecule docking and / or pharmacodynamic parameter evaluation on the 3D conformation and the target protein, and preliminarily selecting potential active molecules or druggable molecules; The eighth module: through a global optimization algorithm, iteratively updating the selected potential active molecules and / or druggable molecules, and outputting better target series of molecules; The ninth module: generating a report of the better target series of molecules.

9. A molecular design and structure optimization device, including a processor and a memory; the memory is used to store a program; the processor executes the program to implement the method according to any one of claims 1-7.

10. A computer-readable storage medium Characterized in that The storage medium stores a program, and the program is executed by a processor to implement the method according to any one of claims 1-7.

Images