A molecular similarity calculation method and device based on molecular structure fingerprint

Abstract

The present invention relates to the field of chemical informatics, a molecular similarity calculation method and device based on molecular structure fingerprint, including: constructing a multi-scale model of the molecule to be tested, determining the atomic type, chemical bond type, and spatial conformation of the molecule to be tested; constructing a molecular graph of the molecule to be tested, determining the molecular characteristics of the molecule to be tested, and constructing the molecular structure fingerprint of the molecule to be tested; analyzing the structural parameters of the molecule to be tested, constructing similar matching items of the molecule to be tested through the structural parameters, calculating the correlation coefficient of the similar matching items, and constructing a cluster association group of the molecule to be tested; determining the maximum similar matching item of the cluster association group, calculating the similarity measure of the maximum similar matching item; determining the matching order of the cluster association group, constructing a similarity matrix of the molecule to be tested, and analyzing the similarity of the molecule to be tested. The present invention can improve the accuracy and timeliness of molecular similarity calculation.

Description

Technical Field

[0001] The present invention relates to the field of chemical informatics, and in particular to a molecular similarity calculation method and device based on molecular structure fingerprint. Background Art

[0002] Molecular similarity calculation refers to the process of evaluating the degree of structural similarity between two or more molecules through quantitative methods. Similarity calculation helps to identify compounds with similar biological activities, predict the properties of compounds, and screen potential drug candidate molecules in drug discovery.

[0003] Currently, the traditional method of calculating molecular similarity mainly relies on simple mathematical metrics to compare molecular features. However, compared with the method based on molecular structure fingerprints, its disadvantage is that it cannot fully capture the complex structure of molecules, resulting in low accuracy and timeliness of analytical similarity evaluation. Summary of the invention

[0004] The present invention provides a molecular similarity calculation method and device based on molecular structure fingerprint, the main purpose of which is to improve the accuracy and timeliness of molecular similarity calculation.

[0005] To achieve the above object, the present invention provides a molecular similarity calculation method based on molecular structure fingerprint, comprising:

[0006] Acquiring structural information of the molecule to be tested, constructing a multi-scale model of the molecule to be tested based on the structural information, and determining the atomic type, chemical bond type, and spatial conformation of the molecule to be tested based on the multi-scale model;

[0007] Based on the atomic type, chemical bond type, and spatial conformation, a molecular graph of the molecule to be tested is constructed; based on the molecular graph and a preset structural GNN model, the molecular characteristics of the molecule to be tested are determined; the molecular fingerprint category of the molecule to be tested is determined in a preset molecular fingerprint library; based on the molecular characteristics and the molecular fingerprint category, a molecular structure fingerprint of the molecule to be tested is constructed;

[0008] Based on the molecular structure fingerprint, the structural parameters of the molecule to be tested are analyzed, similar matching items of the molecule to be tested are constructed through the structural parameters, the correlation coefficients of the similar matching items are calculated, and based on the correlation coefficients and the similar matching items, a cluster association group of the molecule to be tested is constructed;

[0009] Determine the maximum similarity matching item of the cluster association group, and calculate the similarity measure of the maximum similarity matching item;

[0010] The matching order of the cluster association groups is determined by the similarity metric, a similarity matrix of the molecules to be tested is constructed based on the matching order, and the similarity of the molecules to be tested is analyzed based on the similarity matrix.

[0011] Optionally, constructing a multi-scale model of the molecule to be tested based on the structural information comprises:

[0012] The structure information is loaded using an interface function corresponding to a preset OpenBabel tool to obtain loaded structure information;

[0013] converting the loaded structural information into a molecular model;

[0014] Determining a model scale of the molecular model, wherein the model scale includes an electronic level and a molecular level;

[0015] Based on the electronic level, analyzing the electronic structure and electronic energy level of the molecular model;

[0016] Based on the molecular level, determining the interaction force and conformation of the molecular model;

[0017] Constructing an initial multi-scale model of the molecule to be tested through the electronic structure, electronic energy level, interaction force, conformation and molecular model;

[0018] Optimizing the corresponding conformation of the initial multi-scale model to obtain an optimized multi-scale model;

[0019] The conformational stability of the optimized multi-scale model is calculated, and when the conformational stability meets a preset stability threshold, the optimized multi-scale model is used as the multi-scale model of the molecule to be tested.

[0020] Optionally, optimizing the corresponding conformation of the initial multi-scale model to obtain an optimized multi-scale model includes:

[0021] Integrating the initial multi-scale model into a preset GROMACS tool to obtain an initial multi-scale model to be optimized;

[0022] configuring force field parameters of the initial multiscale model to be optimized in the GROMACS tool;

[0023] Based on the force field parameters, the atomic forces of the corresponding conformation of the initial multiscale model are calculated using the following formula:

[0024] ;

[0025] in, represents the atomic forces corresponding to the conformation of the initial multiscale model, represents the attraction constant corresponding to the force field parameters, represents the repulsive force constant corresponding to the force field parameters, represents the hydrogen bond energy parameter corresponding to the force field parameter, represents the hydrogen bond order parameter corresponding to the force field parameter, Indicates the conformational atoms corresponding to the initial multiscale model and atoms The distance Indicates the conformational atoms corresponding to the initial multiscale model and atoms The distance Indicates the conformational atoms corresponding to the initial multiscale model The charge, Indicates the conformational atoms corresponding to the initial multiscale model The charge of

[0026] Based on the atomic forces, the corresponding conformation of the initial multi-scale model is optimized to obtain an optimized multi-scale model.

[0027] Optionally, determining the atomic type, chemical bond type, and spatial conformation of the molecule to be tested based on the multi-scale model includes:

[0028] marking the atomic coordinates of the atoms corresponding to the multiscale model and identifying the atomic masses of the atoms;

[0029] Determining the atomic type of the molecule to be measured based on the atomic coordinates and atomic mass;

[0030] Calculating atomic distances and atomic angles of the atoms using the atomic coordinates;

[0031] Determining the chemical bond type of the molecule to be tested according to the atomic distance and the atomic angle;

[0032] Using the atomic coordinates, constructing the planar structure and three-dimensional configuration of the molecule to be tested;

[0033] The spatial conformation of the molecule to be tested is analyzed according to the planar structure and the three-dimensional configuration.

[0034] Optionally, constructing a molecular graph of the molecule to be tested based on the atomic type, chemical bond type, and spatial conformation includes:

[0035] Based on the atomic type, determining the element symbol of the atom corresponding to the molecule to be tested;

[0036] Marking the connection relationship between the corresponding atoms of the molecule to be tested by the chemical bond type;

[0037] Determining the spatial parameters of the atoms corresponding to the molecule to be tested according to the spatial conformation;

[0038] Using the element symbols as graph nodes of the molecules to be tested, and using the connection relationships as graph edges of the molecules to be tested;

[0039] Based on the graph nodes and graph edges, construct an initial molecular graph of the molecule to be tested;

[0040] Utilizing the spatial parameters to optimize the information of the initial molecular graph, and obtaining an optimized molecular graph of the molecule to be tested;

[0041] Performing a self-loop traversal on the optimized molecular graph to obtain a self-loop traversal set;

[0042] When the self-loop traversal set is empty, the optimized molecular graph is used as the molecular graph of the molecule to be tested.

[0043] Optionally, constructing the molecular structure fingerprint of the molecule to be tested based on the molecular features and the molecular fingerprint category includes:

[0044] Defining a hash function of the molecule to be tested by using the molecular fingerprint category;

[0045] Constructing a hash table of the molecules to be tested;

[0046] Mapping the molecular feature to a corresponding index in a hash table using the hash function to obtain a mapping index of the molecular feature;

[0047] Integrate the mapping indexes to obtain a hash code;

[0048] The molecular structure fingerprint of the molecule to be tested is constructed through the hash code.

[0049] Optionally, mapping the molecular feature to a corresponding index in a hash table using the hash function to obtain a mapping index of the molecular feature includes:

[0050] Based on the hash function, the characteristic hash value of the molecular feature is calculated using the following formula:

[0051] ;

[0052] in, Indicates The feature hash value of each molecular feature, represents a hash function, Indicates the index range of the hash table corresponding to the molecular feature, Represents modulo operation;

[0053] Based on the feature hash value, mapping the molecular feature to a corresponding index in a hash table to obtain an initial mapping index of the molecular feature;

[0054] Calculating a mapping conflict coefficient of the initial mapping index;

[0055] When the mapping conflict coefficient meets a preset conflict threshold, the initial mapping index is used as the mapping index of the molecular feature.

[0056] Optionally, calculating the correlation coefficient of the similar matching items includes:

[0057] constructing an item matrix of the similar matching items;

[0058] Standardizing the project matrix to obtain a standardized project matrix;

[0059] calculating a covariance matrix of the standardized item matrix;

[0060] The covariance matrix is ​​solved using the following formula to obtain eigenvalues ​​and eigenvectors:

[0061] ;

[0062] in, represents the determinant, represents the covariance matrix, represents the identity matrix, represents the eigenvalue, Indicates similar matching items and The covariance of similar matching items, Indicates eigenvalues, Indicates The eigenvector corresponding to the eigenvalue;

[0063] Constructing a score matrix and a loading matrix of the similar matching items through the eigenvalues ​​and eigenvectors;

[0064] Based on the score matrix and the loading matrix, the correlation coefficients of the similar matching items are calculated.

[0065] Optionally, constructing a similarity matrix of the molecules to be tested based on the matching order comprises:

[0066] Based on the matching order, determining the sequence cluster association group of the molecule to be tested;

[0067] Configuring an association group similarity matching weight of the sequence cluster association group;

[0068] Calculate the similarity of the matching items of the sequence cluster association group corresponding to the similar matching items by using the molecular structure fingerprint corresponding to the molecule to be tested;

[0069] Configure the item similarity matching weight of the similarity matching item corresponding to the sequence cluster association group;

[0070] The similarity matrix of the molecule to be tested is constructed by using the association group similarity matching weights, matching item similarities and item similarity matching weights.

[0071] To achieve the above object, the present invention also provides a molecular similarity calculation device based on molecular structure fingerprint, comprising:

[0072] A multi-scale model building module, used to obtain structural information of the molecule to be tested, build a multi-scale model of the molecule to be tested based on the structural information, and determine the atomic type, chemical bond type, and spatial conformation of the molecule to be tested based on the multi-scale model;

[0073] A molecular structure fingerprint construction module is used to construct a molecular graph of the molecule to be tested based on the atomic type, chemical bond type, and spatial conformation, determine the molecular characteristics of the molecule to be tested based on the molecular graph and a preset structural GNN model, determine the molecular fingerprint category of the molecule to be tested in a preset molecular fingerprint library, and construct a molecular structure fingerprint of the molecule to be tested based on the molecular characteristics and the molecular fingerprint category;

[0074] A cluster association group construction module is used to analyze the structural parameters of the molecule to be tested based on the molecular structure fingerprint, construct similar matching items of the molecule to be tested through the structural parameters, calculate the correlation coefficient of the similar matching items, and construct the cluster association group of the molecule to be tested based on the correlation coefficient and the similar matching items;

[0075] A maximum similarity matching item analysis module, used to determine the maximum similarity matching item of the cluster association group and calculate the similarity measure of the maximum similarity matching item;

[0076] The molecular similarity calculation module is used to determine the matching order of the cluster association groups through the similarity metric, construct a similarity matrix of the molecules to be tested based on the matching order, and analyze the similarity of the molecules to be tested based on the similarity matrix.

[0077] In order to solve the above problem, the present invention further provides an electronic device, the electronic device comprising:

[0078] a memory storing at least one instruction; and

[0079] The processor executes the instructions stored in the memory to implement the above-mentioned molecular similarity calculation method based on molecular structure fingerprint.

[0080] In order to solve the above problems, the present invention also provides a computer-readable storage medium, in which at least one instruction is stored. The at least one instruction is executed by a processor in an electronic device to implement the above-mentioned molecular similarity calculation method based on molecular structure fingerprint.

[0081] Based on the structural information, the present invention constructs a multi-scale model of the molecule to be tested, which can obtain accurate molecular information of the molecule, thereby improving the accuracy of the subsequent similarity calculation; based on the molecular graph and the preset structural GNN model, the molecular characteristics of the molecule to be tested are determined to provide a basis for the later structural fingerprint; further, the present invention calculates the correlation coefficient of the similar matching items to analyze the correlation between similar matching items for grouping, and optimizes the efficiency of molecular similarity calculation accordingly. The present invention determines the matching order of the cluster association group through the similarity metric, and based on the matching order, important matching items can be preferentially matched, thereby improving the efficiency of molecular similarity calculation. Further, the present invention constructs a similarity matrix of the molecule to be tested based on the matching order, which can be used as a basis for evaluating molecular similarity. Finally, the present invention analyzes the similarity of the molecule to be tested based on the similarity matrix, which can realize the accurate calculation of the similarity of the molecule to be tested. Therefore, the present invention can improve the accuracy and timeliness of molecular similarity calculation. BRIEF DESCRIPTION OF THE DRAWINGS

[0082] Figure 1 A schematic diagram of a flow chart of a molecular similarity calculation method based on molecular structure fingerprint provided by one embodiment of the present invention;

[0083] Figure 2 A functional module diagram of a molecular similarity calculation device based on molecular structure fingerprint provided by one embodiment of the present invention;

[0084] Figure 3 A schematic diagram of the structure of an electronic device for implementing the molecular similarity calculation method based on molecular structure fingerprint provided by an embodiment of the present invention.

[0085] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0086] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0087] The embodiment of the present application provides a molecular similarity calculation method based on molecular structure fingerprint. The execution subject of the molecular similarity calculation method based on molecular structure fingerprint includes but is not limited to at least one of the electronic devices such as a server, a terminal, etc. that can be configured to execute the method provided in the embodiment of the present application. In other words, the molecular similarity calculation method based on molecular structure fingerprint can be executed by software or hardware installed on a terminal device or a server device, and the software can be a blockchain platform. The server includes but is not limited to: a single server, a server cluster, a cloud server or a cloud server cluster, etc.

[0088] Reference Figure 1 FIG. 1 is a flow chart of a molecular similarity calculation method based on molecular structure fingerprint provided by an embodiment of the present invention. In this embodiment, the molecular similarity calculation method based on molecular structure fingerprint includes:

[0089] S1. Obtain structural information of the molecule to be tested, construct a multi-scale model of the molecule to be tested based on the structural information, and determine the atomic type, chemical bond type, and spatial conformation of the molecule to be tested based on the multi-scale model.

[0090] It should be explained that the structural information refers to raw data based on experiments or theoretical calculations, such as molecular formula, molecular structure formula, molecular structure diagram and other data.

[0091] The present invention constructs a multi-scale model of the molecule to be tested based on the structural information to obtain accurate molecular information of the molecule, thereby improving the accuracy of subsequent similarity calculations.

[0092] In detail, constructing a multi-scale model of the molecule to be tested based on the structural information includes:

[0093] The structure information is loaded using an interface function corresponding to a preset OpenBabel tool to obtain loaded structure information;

[0094] converting the loaded structural information into a molecular model;

[0095] Determining a model scale of the molecular model, wherein the model scale includes an electronic level and a molecular level;

[0096] Based on the electronic level, analyzing the electronic structure and electronic energy level of the molecular model;

[0097] Based on the molecular level, determining the interaction force and conformation of the molecular model;

[0098] Constructing an initial multi-scale model of the molecule to be tested through the electronic structure, electronic energy level, interaction force, conformation and molecular model;

[0099] Optimizing the corresponding conformation of the initial multi-scale model to obtain an optimized multi-scale model;

[0100] The conformational stability of the optimized multi-scale model is calculated, and when the conformational stability meets a preset stability threshold, the optimized multi-scale model is used as the multi-scale model of the molecule to be tested.

[0101] Among them, the OpenBabel tool refers to an open source chemical informatics tool that can process molecular structure formats, the interface function refers to a series of interface functions provided by OpenBabel, which allow users to load, convert, calculate and save structural information, etc., the loading of structural information refers to the use of OpenBabel's interface functions to load structural information from files (such as .mol2, .pdb, .sdf, etc.) or strings (such as SMILES, InChI), the molecular model refers to the representation of the molecular structure stored in the computer, including atomic coordinates, chemical bonds, atomic types, charges and other information, the electronic level refers to the layer for studying the electronic structure and electronic energy level in the molecular model, the molecular level refers to the layer for studying the molecular interaction force in the molecular model, and the electronic structure refers to the electrons in atoms or molecules. distribution and arrangement, the electronic energy level refers to the energy state of electrons in atoms or molecules, the interaction force refers to the force between atoms due to the overlap of electron clouds, the initial multi-scale model is constructed based on structural information, and includes a model that initially combines the calculation results of the electronic level (QM) and the molecular level (MM), the optimized multi-scale model refers to a model of the lowest energy stable conformation found by a conformation optimization algorithm (such as MM-PBSA, MM-GBSA, etc.) on the basis of the initial multi-scale model, the conformational stability refers to the stable conformation index of a molecule under certain conditions, the stability threshold refers to a preset threshold of an energy or conformational evaluation index, which is used to judge whether the conformational stability meets the requirements, the conformation refers to the three-dimensional arrangement of molecules in space, and the multi-scale model refers to a calculation method that combines different scale models such as the electronic level and the molecular level.

[0102] Furthermore, the optimizing the corresponding conformation of the initial multi-scale model to obtain an optimized multi-scale model includes:

[0103] Integrating the initial multi-scale model into a preset GROMACS tool to obtain an initial multi-scale model to be optimized;

[0104] configuring force field parameters of the initial multiscale model to be optimized in the GROMACS tool;

[0105] Based on the force field parameters, the atomic forces of the corresponding conformation of the initial multiscale model are calculated using the following formula:

[0106] ;

[0107] in, represents the atomic forces corresponding to the conformation of the initial multiscale model, represents the attraction constant corresponding to the force field parameters, represents the repulsive force constant corresponding to the force field parameters, represents the hydrogen bond energy parameter corresponding to the force field parameter, represents the hydrogen bond order parameter corresponding to the force field parameter, Indicates the conformational atoms corresponding to the initial multiscale model and atoms The distance Indicates the conformational atoms corresponding to the initial multiscale model and atoms The distance Indicates the conformational atoms corresponding to the initial multiscale model The charge, Indicates the conformational atoms corresponding to the initial multiscale model The charge of

[0108] Based on the atomic forces, the corresponding conformation of the initial multi-scale model is optimized to obtain an optimized multi-scale model.

[0109] Among them, the GROMACS tool refers to software or platform for molecular dynamics simulation and conformation optimization, and the force field parameters refer to empirical parameters used to describe the intermolecular interaction forces. These parameters are determined based on a large amount of experimental data and theoretical derivation, and are used to simulate non-covalent interactions such as van der Waals forces, electrostatic interactions, and hydrogen bonds. The atomic forces refer to and are exemplified by forces generated between atoms due to the overlap of electron clouds, including van der Waals forces, electrostatic interactions, hydrogen bonds, etc. The attraction constant refers to the description of the van der Waals attraction between atoms in a molecule, the repulsion constant refers to the description of the van der Waals repulsion between atoms in a molecule, the hydrogen bond energy parameter refers to the parameter describing the strength and stability of hydrogen bonds, and the hydrogen bond order parameter refers to the parameter describing the bond order and strength of hydrogen bonds.

[0110] Based on the multi-scale model, determining the atomic type, chemical bond type, and spatial conformation of the molecule to be tested can serve as a basis for constructing a high-dimensional molecular fingerprint in the later stage.

[0111] In detail, the determining of the atomic type, chemical bond type, and spatial conformation of the molecule to be tested based on the multi-scale model includes:

[0112] marking the atomic coordinates of the atoms corresponding to the multiscale model and identifying the atomic masses of the atoms;

[0113] Determining the atomic type of the molecule to be measured based on the atomic coordinates and atomic mass;

[0114] Calculating atomic distances and atomic angles of the atoms using the atomic coordinates;

[0115] Determining the chemical bond type of the molecule to be tested according to the atomic distance and the atomic angle;

[0116] Using the atomic coordinates, constructing the planar structure and three-dimensional configuration of the molecule to be tested;

[0117] The spatial conformation of the molecule to be tested is analyzed according to the planar structure and the three-dimensional configuration.

[0118] Among them, the atomic coordinates refer to the three-dimensional spatial position of each atom in the molecule, the atomic mass refers to the mass of each atom in the molecule, the atomic type refers to the element type of each atom in the molecule, such as carbon (C), hydrogen (H), nitrogen (N), oxygen (O), etc., the atomic distance refers to the spatial distance between any two atoms in the molecule, the atomic angle refers to the angle between any two atoms in the plane formed by any three atoms in the molecule, the chemical bond type refers to the type of chemical bond between different atoms in the molecule, such as single bond, double bond, triple bond, etc., the planar structure refers to the arrangement of certain atoms in the molecule on the same plane, the stereo configuration refers to the arrangement of atoms in the molecule in three-dimensional space, and the spatial conformation refers to the actual shape and arrangement of the molecule in three-dimensional space.

[0119] S2. Based on the atomic type, chemical bond type, and spatial conformation, a molecular graph of the molecule to be tested is constructed; based on the molecular graph and a preset structural GNN model, the molecular characteristics of the molecule to be tested are determined; the molecular fingerprint category of the molecule to be tested is determined in a preset molecular fingerprint library; based on the molecular characteristics and the molecular fingerprint category, a molecular structure fingerprint of the molecule to be tested is constructed.

[0120] Based on the atomic type, chemical bond type, and spatial conformation, a molecular graph of the molecule to be tested is constructed, which can be used for subsequent feature analysis of a graph neural network.

[0121] In detail, the molecular graph of the molecule to be tested is constructed based on the atomic type, chemical bond type, and spatial conformation, including:

[0122] Based on the atomic type, determining the element symbol of the atom corresponding to the molecule to be tested;

[0123] Marking the connection relationship between the corresponding atoms of the molecule to be tested by the chemical bond type;

[0124] Determining the spatial parameters of the atoms corresponding to the molecule to be tested according to the spatial conformation;

[0125] Using the element symbols as graph nodes of the molecules to be tested, and using the connection relationships as graph edges of the molecules to be tested;

[0126] Based on the graph nodes and graph edges, construct an initial molecular graph of the molecule to be tested;

[0127] Utilizing the spatial parameters to optimize the information of the initial molecular graph, and obtaining an optimized molecular graph of the molecule to be tested;

[0128] Performing a self-loop traversal on the optimized molecular graph to obtain a self-loop traversal set;

[0129] When the self-loop traversal set is empty, the optimized molecular graph is used as the molecular graph of the molecule to be tested.

[0130] Among them, the element symbol refers to the symbol of each element in the periodic table of chemical elements, such as H (hydrogen), C (carbon), N (nitrogen), O (oxygen), etc., the connection relationship refers to the type of chemical bonds between different atoms in the molecule, such as single bond, double bond, triple bond, etc., the spatial parameter refers to the position information of the atoms in the molecule in three-dimensional space, the graph node refers to the basic unit in the graph, usually representing an atom, the graph edge refers to the line connecting two graph nodes, usually representing a chemical bond, the initial molecular graph refers to a molecular graph constructed based on atom type, chemical bond type and spatial conformation, the self-loop traversal set refers to all self-loops found during the traversal of the graph (i.e., edges connecting two identical nodes), and the molecular graph refers to a graph representation describing the molecular structure.

[0131] Furthermore, the self-loop traversal of the optimized molecular graph to obtain the self-loop traversal set can be achieved through a traversal function. After traversing all edges, it is checked whether the self-loop traversal set is empty. If the self-loop traversal set is empty, the graph does not have a self-loop. If the self-loop traversal set is not empty, the graph does have a self-loop.

[0132] Based on the molecular graph and the preset structural GNN model, determining the molecular characteristics of the molecule to be tested can provide a basis for the subsequent structural fingerprint.

[0133] In detail, the structural GNN model (Graph Neural Network model) is a deep learning model. In molecular informatics, the structural GNN model is used to extract molecular features from a molecular graph (a graph consisting of atomic nodes and chemical bond edges). The molecular features refer to a series of characteristics extracted from a molecule, such as the combination of atomic types, the type and length of chemical bonds, the topological structure of the molecule, the interaction between molecules, and other characteristics.

[0134] In detail, the molecular fingerprint library includes fingerprint types such as MACCS keys, ECFP (Extended Connectivity Fingerprints), FCFP (Fixed-size Circular Fingerprints), etc. Each fingerprint type has its own specific hash function.

[0135] The present invention constructs a molecular structure fingerprint of the molecule to be tested based on the molecular characteristics, which can comprehensively characterize the structural characteristics of the molecule and improve the efficiency and accuracy of molecular similarity calculation.

[0136] In detail, the molecular structure fingerprint of the molecule to be tested is constructed based on the molecular features and the molecular fingerprint category, including:

[0137] Defining a hash function of the molecule to be tested by using the molecular fingerprint category;

[0138] Constructing a hash table of the molecules to be tested;

[0139] Mapping the molecular feature to a corresponding index in a hash table using the hash function to obtain a mapping index of the molecular feature;

[0140] Integrate the mapping indexes to obtain a hash code;

[0141] The molecular structure fingerprint of the molecule to be tested is constructed through the hash code.

[0142] The hash function refers to a function that maps input data to a fixed-size output (hash value). For example, the hash function may convert a feature string (such as "C", "H") into an integer through a hash algorithm (such as MD5). The hash table refers to a data structure for storing key-value pairs, where the key is mapped to an index in the table through a hash function, and the mapping index is an index that maps a molecular feature to a hash table through a hash function. Each feature has a unique mapping index, and the hash code refers to combining the mapping indexes of all features into a coding vector. The molecular structure fingerprint refers to a hash code based on molecular features, which converts the complex chemical structure information of the molecule into a digital representation that can be processed by a machine learning model.

[0143] Furthermore, the using of the hash function to map the molecular feature to a corresponding index in a hash table to obtain a mapping index of the molecular feature includes:

[0144] Based on the hash function, the characteristic hash value of the molecular feature is calculated using the following formula:

[0145] ;

[0146] in, Indicates The feature hash value of each molecular feature, represents a hash function, Indicates the index range of the hash table corresponding to the molecular feature, Represents modulo operation;

[0147] Based on the feature hash value, mapping the molecular feature to a corresponding index in a hash table to obtain an initial mapping index of the molecular feature;

[0148] Calculating a mapping conflict coefficient of the initial mapping index;

[0149] When the mapping conflict coefficient meets a preset conflict threshold, the initial mapping index is used as the mapping index of the molecular feature.

[0150] Among them, the feature hash value refers to the hash result of the feature vector calculated by the hash function, the index range refers to the set of available indexes in the hash table, the modulo operation refers to dividing the hash value by the size of the hash table and then taking the remainder, the initial mapping index refers to the hash table index obtained based on the feature hash value and the modulo operation, and the mapping conflict coefficient refers to the frequency at which multiple feature vectors are mapped to the same index in the hash table.

[0151] Furthermore, the calculating of the mapping conflict coefficient of the initial mapping index;

[0152] When the mapping conflict coefficient meets the preset conflict threshold, the initial mapping index is used as the mapping index of the molecular feature. When the mapping conflict coefficient does not meet the preset conflict threshold, it can be solved by using open addressing method, chain address method and other methods.

[0153] S3. Analyze the structural parameters of the molecule to be tested based on the molecular structure fingerprint, construct similar matching items of the molecule to be tested through the structural parameters, calculate the correlation coefficient of the similar matching items, and construct a cluster association group of the molecule to be tested based on the correlation coefficient and the similar matching items.

[0154] It should be explained that the structural parameters include parameters such as topological structure, chemical properties, three-dimensional space information, etc. The similarity matching items include items such as atomic type matching, chemical bond type matching, functional group similarity, ring structure similarity, and spatial conformation similarity.

[0155] The present invention constructs similar matching items of the molecule to be tested through the structural parameters for subsequent molecular similarity search and analysis. In detail, the similar matching items can be implemented by screening out key structural parameters from the structural parameters and designing matching items for the key structural parameters. For example, when the key structural parameter is a topological structure, the similar matching item can be a ring structure similarity.

[0156] The present invention calculates the correlation coefficient of the similar matching items to analyze the correlation between the similar matching items for grouping, and optimizes the molecular similarity calculation efficiency accordingly.

[0157] In detail, the calculation of the correlation coefficient of the similar matching items includes:

[0158] constructing an item matrix of the similar matching items;

[0159] Standardizing the project matrix to obtain a standardized project matrix;

[0160] calculating a covariance matrix of the standardized item matrix;

[0161] The covariance matrix is ​​solved using the following formula to obtain eigenvalues ​​and eigenvectors:

[0162] ;

[0163] in, represents the determinant, represents the covariance matrix, represents the identity matrix, represents the eigenvalue, Indicates similar matching items and The covariance of similar matching items, Indicates eigenvalues, Indicates The eigenvector corresponding to the eigenvalue;

[0164] Constructing a score matrix and a loading matrix of the similar matching items through the eigenvalues ​​and eigenvectors;

[0165] Based on the score matrix and the loading matrix, the correlation coefficients of the similar matching items are calculated.

[0166] Among them, the project matrix refers to a two-dimensional data structure, in which each row represents a molecule to be tested, and each column represents a similar matching item (such as atomic type matching, chemical bond type matching, etc.), the standardized project matrix refers to a matrix obtained by standardizing each similar matching item in the project matrix, the covariance matrix refers to a matrix describing the covariance between similar matching items, the eigenvalue refers to the value obtained by solving the covariance matrix, the unit matrix refers to a matrix in which all elements on the main diagonal are 1, the eigenvector refers to a vector corresponding to the eigenvalue, which represents the distribution of data in the direction corresponding to the eigenvalue, the score matrix refers to that obtained by multiplying the standardized project matrix by the eigenvector, the load matrix refers to the transpose of the eigenvector, which shows the relationship between the original similar matching items and the principal component, and the correlation coefficient refers to the degree of correlation between similar matching items.

[0167] Furthermore, the standardization of the project matrix to obtain a standardized project matrix can be achieved through technologies such as the stats package in the R language.

[0168] It should be explained that the cluster association group refers to an association group that integrates similar matching items with high association.

[0169] S4. Determine the maximum similarity matching item of the cluster association group, and calculate the similarity measure of the maximum similarity matching item.

[0170] It should be explained that the maximum similarity matching item refers to the matching item with the greatest influence in the molecular similarity comparison process within the cluster association group.

[0171] Exemplarily, the cluster association group may be an atom type cluster association group, and the maximum similarity matching items may be molecules having the same atom type, such as two molecules both containing four carbon atoms.

[0172] The cluster association group may be a chemical bond type cluster association group, and the maximum similarity matching items may be molecules with the same chemical bond type, such as two molecules both containing two carbon-carbon double bonds.

[0173] The similarity measure refers to the similarity score in the maximum similarity matching item. In detail, the similarity measure can be calculated using the Tanimoto coefficient.

[0174] S5. Determine the matching order of the cluster association groups by the similarity metric, construct a similarity matrix of the molecules to be tested based on the matching order, and analyze the similarity of the molecules to be tested based on the similarity matrix.

[0175] The present invention determines the matching order of the cluster association group through the similarity metric, and based on the matching order, important matching items can be matched first, thereby improving the efficiency of molecular similarity calculation. In detail, the matching order of the cluster association group can be sorted by the similarity metric, and the higher the similarity metric, the higher the matching order.

[0176] The present invention constructs a similarity matrix of the molecules to be tested based on the matching order, which can be used as a basis for evaluating molecular similarity.

[0177] In detail, the similarity matrix of the molecules to be tested is constructed based on the matching order, including:

[0178] Based on the matching order, determining the sequence cluster association group of the molecule to be tested;

[0179] Configuring an association group similarity matching weight of the sequence cluster association group;

[0180] Calculate the similarity of the matching items of the sequence cluster association group corresponding to the similar matching items by using the molecular structure fingerprint corresponding to the molecule to be tested;

[0181] Configure the item similarity matching weight of the similarity matching item corresponding to the sequence cluster association group;

[0182] The similarity matrix of the molecule to be tested is constructed by using the association group similarity matching weights, matching item similarities and item similarity matching weights.

[0183] Among them, the sequence cluster association group refers to the association group arranged in matching order, the association group similarity matching weight refers to the weight assigned to each sequence cluster association group, reflecting the importance of the cluster in the overall similarity evaluation, the matching item similarity refers to the similarity between the molecule to be tested and the molecule in each sequence cluster association group in a specific similarity matching item, the item similarity matching weight refers to the weight assigned to each similarity matching item, reflecting the importance of the matching item in the overall similarity evaluation, and the similarity matrix refers to a matrix containing the matching item similarities and weights of similar matching items between the molecules to be tested.

[0184] Exemplarily, there are two molecules to be tested, A and B. When the sequence cluster association group is an atom type cluster association group and a chemical bond type cluster association group, the association group similarity matching weights of the atom type cluster association group and the chemical bond type cluster association group are 0.6 and 0.4 respectively, the atom type cluster association group includes a carbon (C) atom matching item and a hydrogen (H) matching item, and the item similarity matching weights of the carbon (C) atom matching item and the hydrogen (H) matching item are 0.3 and 0.7 respectively, the chemical bond type cluster association group includes a single bond matching item and a cyclic bond matching item, and the item similarity matching weights of the single bond matching item and the cyclic bond matching item are 0.6 and 0.4 respectively; the matching item similarities of the carbon (C) atom matching item, the hydrogen (H) matching item, the single bond matching item and the cyclic bond matching item are 0.2, 0.4, 0.5, and 0.1 respectively, that is, the similarity matrix of the two molecules to be tested, A and B, is expressed as:

[0185]

[0186] Finally, the present invention analyzes the similarity of the molecules to be tested based on the similarity matrix and can achieve accurate calculation of the similarity of the molecules to be tested.

[0187] Exemplarily, when the similarity matrix of the two molecules to be tested, A and B, is expressed as:

[0188]

[0189] That is, the similarity between the two molecules to be tested, A and B, is:

[0190] .

[0191] Based on the structural information, the present invention constructs a multi-scale model of the molecule to be tested, which can obtain accurate molecular information of the molecule, thereby improving the accuracy of the subsequent similarity calculation; based on the molecular graph and the preset structural GNN model, the molecular characteristics of the molecule to be tested are determined to provide a basis for the later structural fingerprint; further, the present invention calculates the correlation coefficient of the similar matching items to analyze the correlation between similar matching items for grouping, and optimizes the efficiency of molecular similarity calculation accordingly. The present invention determines the matching order of the cluster association group through the similarity metric, and based on the matching order, important matching items can be preferentially matched, thereby improving the efficiency of molecular similarity calculation. Further, the present invention constructs a similarity matrix of the molecule to be tested based on the matching order, which can be used as a basis for evaluating molecular similarity. Finally, the present invention analyzes the similarity of the molecule to be tested based on the similarity matrix, which can realize the accurate calculation of the similarity of the molecule to be tested. Therefore, the present invention can improve the accuracy and timeliness of molecular similarity calculation.

[0192] like Figure 2 , which is a functional module diagram of a molecular similarity calculation device based on molecular structure fingerprint provided by one embodiment of the present invention.

[0193] The molecular similarity calculation device 100 based on molecular structure fingerprint of the present invention can be installed in an electronic device. According to the functions implemented, the molecular similarity calculation device 100 based on molecular structure fingerprint can include a multi-scale model construction module 101, a molecular structure fingerprint construction module 102, a cluster association group construction module 103, a maximum similarity matching item analysis module 104 and a molecular similarity calculation module 105. The module of the present invention can also be called a unit, which refers to a series of computer program segments that can be executed by an electronic device processor and can complete fixed functions, which are stored in the memory of the electronic device.

[0194] The multi-scale model building module 101 is used to obtain structural information of the molecule to be tested, build a multi-scale model of the molecule to be tested based on the structural information, and determine the atomic type, chemical bond type, and spatial conformation of the molecule to be tested based on the multi-scale model;

[0195] The molecular structure fingerprint construction module 102 is used to construct a molecular graph of the molecule to be tested based on the atomic type, chemical bond type, and spatial conformation, determine the molecular characteristics of the molecule to be tested based on the molecular graph and a preset structural GNN model, determine the molecular fingerprint category of the molecule to be tested in a preset molecular fingerprint library, and construct a molecular structure fingerprint of the molecule to be tested based on the molecular characteristics and the molecular fingerprint category;

[0196] The cluster association group construction module 103 is used to analyze the structural parameters of the molecule to be tested based on the molecular structure fingerprint, construct similar matching items of the molecule to be tested through the structural parameters, calculate the correlation coefficient of the similar matching items, and construct the cluster association group of the molecule to be tested based on the correlation coefficient and the similar matching items;

[0197] The maximum similarity matching item analysis module 104 is used to determine the maximum similarity matching item of the cluster association group and calculate the similarity measure of the maximum similarity matching item;

[0198] The molecular similarity calculation module 105 is used to determine the matching order of the cluster association groups through the similarity metric, construct a similarity matrix of the molecules to be tested based on the matching order, and analyze the similarity of the molecules to be tested based on the similarity matrix.

[0199] In detail, each module in the molecular similarity calculation device 100 based on molecular structure fingerprint in the embodiment of the present invention is used in the same manner as described above. Figure 1The molecular similarity calculation method based on molecular structure fingerprint described in the present invention is the same technical means and can produce the same technical effect, which will not be repeated here.

[0200] like Figure 3 FIG. 1 is a schematic diagram of the structure of an electronic device for implementing a molecular similarity calculation method based on molecular structure fingerprints provided by an embodiment of the present invention.

[0201] The electronic device 1 may include a processor 10, a memory 11 and a bus 12, and may also include a computer program stored in the memory 11 and executable on the processor 10, such as a molecular similarity calculation method program based on molecular structure fingerprints.

[0202] Wherein, the memory 11 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, mobile hard disk, multimedia card, card-type memory (for example: SD or DX memory, etc.), magnetic memory, disk, optical disk, etc. In some embodiments, the memory 11 can be an internal storage unit of the electronic device 1, such as a mobile hard disk of the electronic device 1. In other embodiments, the memory 11 can also be an external storage device of the electronic device 1, such as a plug-in mobile hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (SecureDigital, SD) card, a flash card (Flash Card), etc. equipped on the electronic device 1. Further, the memory 11 also includes an internal storage unit of the electronic device 1 and an external storage device. The memory 11 can not only be used to store application software and various types of data installed in the electronic device 1, such as the code of the molecular similarity calculation method program based on molecular structure fingerprints, but also can be used to temporarily store data that has been output or is to be output.

[0203] The processor 10 may be composed of an integrated circuit in some embodiments, for example, a single packaged integrated circuit, or a plurality of packaged integrated circuits with the same or different functions, including one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processors, and combinations of various control chips, etc. The processor 10 is the control core (Control Unit) of the electronic device, and uses various interfaces and lines to connect various components of the entire electronic device, and executes or executes programs or modules stored in the memory 11 (such as a molecular similarity calculation method program based on molecular structure fingerprints, etc.), and calls data stored in the memory 11 to execute various functions of the electronic device 1 and process data.

[0204] The bus 12 may be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus, etc. The bus 12 may be divided into an address bus, a data bus, a control bus, etc. The bus 12 is configured to realize connection and communication between the memory 11 and at least one processor 10, etc.

[0205] Figure 3 Only an electronic device with components is shown, and those skilled in the art will understand that Figure 3 The structure shown does not constitute a limitation on the electronic device 1, and may include fewer or more components than shown in the figure, or combine certain components, or arrange the components differently.

[0206] For example, although not shown, the electronic device 1 may also include a power source (such as a battery) for supplying power to various components. Preferably, the power source may be logically connected to the at least one processor 10 through a power management device, so that the power management device can realize functions such as charging management, discharging management, and power consumption management. The power source may also include one or more DC or AC power sources, recharging devices, power failure detection circuits, power converters or inverters, power status indicators, and other arbitrary components. The electronic device 1 may also include a variety of sensors, Bluetooth modules, Wi-Fi modules, etc., which will not be repeated here.

[0207] Furthermore, the electronic device 1 may also include a network interface. Optionally, the network interface may include a wired interface and / or a wireless interface (such as a WI-FI interface, a Bluetooth interface, etc.), which is generally used to establish a communication connection between the electronic device 1 and other electronic devices.

[0208] Optionally, the electronic device 1 may further include a user interface, which may be a display, an input unit (such as a keyboard), or a standard wired interface or a wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, and an OLED (Organic Light-Emitting Diode) touch device. The display may also be appropriately referred to as a display screen or a display unit, which is used to display information processed in the electronic device 1 and to display a visual user interface.

[0209] It should be understood that the embodiment is for illustration only and the scope of the patent application is not limited to this structure.

[0210] The molecular similarity calculation method program based on molecular structure fingerprint stored in the memory 11 of the electronic device 1 is a combination of multiple instructions. When running in the processor 10, it can achieve:

[0211] Acquiring structural information of the molecule to be tested, constructing a multi-scale model of the molecule to be tested based on the structural information, and determining the atomic type, chemical bond type, and spatial conformation of the molecule to be tested based on the multi-scale model;

[0212] Based on the atomic type, chemical bond type, and spatial conformation, a molecular graph of the molecule to be tested is constructed; based on the molecular graph and a preset structural GNN model, the molecular characteristics of the molecule to be tested are determined; the molecular fingerprint category of the molecule to be tested is determined in a preset molecular fingerprint library; based on the molecular characteristics and the molecular fingerprint category, a molecular structure fingerprint of the molecule to be tested is constructed;

[0213] Based on the molecular structure fingerprint, the structural parameters of the molecule to be tested are analyzed, similar matching items of the molecule to be tested are constructed through the structural parameters, the correlation coefficients of the similar matching items are calculated, and based on the correlation coefficients and the similar matching items, a cluster association group of the molecule to be tested is constructed;

[0214] Determine the maximum similarity matching item of the cluster association group, and calculate the similarity measure of the maximum similarity matching item;

[0215] The matching order of the cluster association groups is determined by the similarity metric, a similarity matrix of the molecules to be tested is constructed based on the matching order, and the similarity of the molecules to be tested is analyzed based on the similarity matrix.

[0216] Furthermore, if the module / unit integrated in the electronic device 1 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. The computer-readable storage medium can be volatile or non-volatile. For example, the computer-readable medium can include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a mobile hard disk, a magnetic disk, an optical disk, a computer memory, and a read-only memory (ROM).

[0217] The present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor of an electronic device, the computer program can implement:

[0218] Acquiring structural information of the molecule to be tested, constructing a multi-scale model of the molecule to be tested based on the structural information, and determining the atomic type, chemical bond type, and spatial conformation of the molecule to be tested based on the multi-scale model;

[0219] Based on the atomic type, chemical bond type, and spatial conformation, a molecular graph of the molecule to be tested is constructed; based on the molecular graph and a preset structural GNN model, the molecular characteristics of the molecule to be tested are determined; the molecular fingerprint category of the molecule to be tested is determined in a preset molecular fingerprint library; based on the molecular characteristics and the molecular fingerprint category, a molecular structure fingerprint of the molecule to be tested is constructed;

[0220] Based on the molecular structure fingerprint, the structural parameters of the molecule to be tested are analyzed, similar matching items of the molecule to be tested are constructed through the structural parameters, the correlation coefficients of the similar matching items are calculated, and based on the correlation coefficients and the similar matching items, a cluster association group of the molecule to be tested is constructed;

[0221] Determine the maximum similarity matching item of the cluster association group, and calculate the similarity measure of the maximum similarity matching item;

[0222] The matching order of the cluster association groups is determined by the similarity metric, a similarity matrix of the molecules to be tested is constructed based on the matching order, and the similarity of the molecules to be tested is analyzed based on the similarity matrix.

[0223] In the several embodiments provided by the present invention, it should be understood that the disclosed devices, apparatuses and methods can be implemented in other ways. For example, the device embodiments described above are only illustrative, and actual implementation may have other division methods.

[0224] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0225] In addition, each functional module in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of hardware plus software functional modules.

[0226] It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0227] In addition, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices stated in a device claim can also be implemented by one unit or device through software or hardware. The second and other words are used to indicate names, but not to indicate any particular order.

[0228] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. A molecular similarity calculation method based on molecular structure fingerprint, characterized in that: The method comprises: Acquiring structural information of the molecule to be tested, constructing a multi-scale model of the molecule to be tested based on the structural information, and determining the atomic type, chemical bond type, and spatial conformation of the molecule to be tested based on the multi-scale model; Based on the atomic type, chemical bond type, and spatial conformation, a molecular graph of the molecule to be tested is constructed; based on the molecular graph and a preset structural GNN model, the molecular characteristics of the molecule to be tested are determined; the molecular fingerprint category of the molecule to be tested is determined in a preset molecular fingerprint library; based on the molecular characteristics and the molecular fingerprint category, a molecular structure fingerprint of the molecule to be tested is constructed; Based on the molecular structure fingerprint, the structural parameters of the molecule to be tested are analyzed, similar matching items of the molecule to be tested are constructed through the structural parameters, the correlation coefficients of the similar matching items are calculated, and based on the correlation coefficients and the similar matching items, a cluster association group of the molecule to be tested is constructed; Determine the maximum similarity matching item of the cluster association group, and calculate the similarity measure of the maximum similarity matching item; The matching order of the cluster association groups is determined by the similarity metric, a similarity matrix of the molecules to be tested is constructed based on the matching order, and the similarity of the molecules to be tested is analyzed based on the similarity matrix.

2. The molecular similarity calculation method based on molecular structure fingerprint according to claim 1, characterized in that: The step of constructing a multi-scale model of the molecule to be tested based on the structural information comprises: The structure information is loaded using an interface function corresponding to a preset OpenBabel tool to obtain loaded structure information; converting the loaded structural information into a molecular model; Determining a model scale of the molecular model, wherein the model scale includes an electronic level and a molecular level; Based on the electronic level, analyzing the electronic structure and electronic energy level of the molecular model; Based on the molecular level, determining the interaction force and conformation of the molecular model; Constructing an initial multi-scale model of the molecule to be tested through the electronic structure, electronic energy level, interaction force, conformation and molecular model; Optimizing the corresponding conformation of the initial multi-scale model to obtain an optimized multi-scale model; The conformational stability of the optimized multi-scale model is calculated, and when the conformational stability meets a preset stability threshold, the optimized multi-scale model is used as the multi-scale model of the molecule to be tested.

3. The molecular similarity calculation method based on molecular structure fingerprint according to claim 2, characterized in that: The step of optimizing the corresponding conformation of the initial multi-scale model to obtain an optimized multi-scale model comprises: Integrating the initial multi-scale model into a preset GROMACS tool to obtain an initial multi-scale model to be optimized; configuring force field parameters of the initial multiscale model to be optimized in the GROMACS tool; Based on the force field parameters, the atomic forces of the corresponding conformation of the initial multiscale model are calculated using the following formula: ;in, represents the atomic forces corresponding to the conformation of the initial multiscale model, represents the attraction constant corresponding to the force field parameters, represents the repulsive force constant corresponding to the force field parameters, represents the hydrogen bond energy parameter corresponding to the force field parameter, represents the hydrogen bond order parameter corresponding to the force field parameter, Indicates the conformational atoms corresponding to the initial multiscale model and atoms The distance Indicates the conformational atoms corresponding to the initial multiscale model and atoms The distance Indicates the conformational atoms corresponding to the initial multiscale model The charge, Indicates the conformational atoms corresponding to the initial multiscale model The charge of Based on the atomic forces, the corresponding conformation of the initial multi-scale model is optimized to obtain an optimized multi-scale model.

4. The molecular similarity calculation method based on molecular structure fingerprint according to claim 3, characterized in that: Determining the atomic type, chemical bond type, and spatial conformation of the molecule to be tested based on the multi-scale model includes: marking the atomic coordinates of the atoms corresponding to the multiscale model and identifying the atomic masses of the atoms; Determining the atomic type of the molecule to be measured based on the atomic coordinates and atomic mass; Calculating atomic distances and atomic angles of the atoms using the atomic coordinates; Determining the chemical bond type of the molecule to be tested according to the atomic distance and the atomic angle; Using the atomic coordinates, constructing the planar structure and three-dimensional configuration of the molecule to be tested; The spatial conformation of the molecule to be tested is analyzed according to the planar structure and the three-dimensional configuration.

5. The molecular similarity calculation method based on molecular structure fingerprint according to claim 4, characterized in that: The step of constructing a molecular graph of the molecule to be tested based on the atomic type, chemical bond type, and spatial conformation includes: Based on the atomic type, determining the element symbol of the atom corresponding to the molecule to be tested; Marking the connection relationship between the corresponding atoms of the molecule to be tested by the chemical bond type; Determining the spatial parameters of the atoms corresponding to the molecule to be tested according to the spatial conformation; Using the element symbols as graph nodes of the molecules to be tested, and using the connection relationships as graph edges of the molecules to be tested; Based on the graph nodes and graph edges, construct an initial molecular graph of the molecule to be tested; Utilizing the spatial parameters to optimize the information of the initial molecular graph, and obtaining an optimized molecular graph of the molecule to be tested; Performing a self-loop traversal on the optimized molecular graph to obtain a self-loop traversal set; When the self-loop traversal set is empty, the optimized molecular graph is used as the molecular graph of the molecule to be tested.

6. The molecular similarity calculation method based on molecular structure fingerprint according to claim 5, characterized in that: The step of constructing a molecular structure fingerprint of the molecule to be tested based on the molecular features and the molecular fingerprint category comprises: Defining a hash function of the molecule to be tested by using the molecular fingerprint category; Constructing a hash table of the molecules to be tested; Mapping the molecular feature to a corresponding index in a hash table using the hash function to obtain a mapping index of the molecular feature; Integrate the mapping indexes to obtain a hash code; The molecular structure fingerprint of the molecule to be tested is constructed through the hash code.

7. The molecular similarity calculation method based on molecular structure fingerprint according to claim 6, characterized in that: The using the hash function to map the molecular feature to a corresponding index in a hash table to obtain a mapping index of the molecular feature includes: Based on the hash function, the characteristic hash value of the molecular feature is calculated using the following formula: ;in, Indicates The feature hash value of each molecular feature, represents a hash function, Indicates the index range of the hash table corresponding to the molecular feature, Represents modulo operation; Based on the feature hash value, mapping the molecular feature to a corresponding index in a hash table to obtain an initial mapping index of the molecular feature; Calculating a mapping conflict coefficient of the initial mapping index; When the mapping conflict coefficient meets a preset conflict threshold, the initial mapping index is used as the mapping index of the molecular feature.

8. The molecular similarity calculation method based on molecular structure fingerprint according to claim 7, characterized in that: The calculating the correlation coefficient of the similar matching items includes: constructing an item matrix of the similar matching items; Standardizing the project matrix to obtain a standardized project matrix; calculating a covariance matrix of the standardized item matrix; The covariance matrix is ​​solved using the following formula to obtain eigenvalues ​​and eigenvectors: ;in, represents the determinant, represents the covariance matrix, represents the identity matrix, represents the eigenvalue, Indicates similar matching items and The covariance of similar matching items, Indicates eigenvalues, Indicates The eigenvector corresponding to the eigenvalue; Constructing a score matrix and a loading matrix of the similar matching items through the eigenvalues ​​and eigenvectors; Based on the score matrix and the loading matrix, the correlation coefficients of the similar matching items are calculated.

9. The molecular similarity calculation method based on molecular structure fingerprint according to claim 8, characterized in that: The constructing of a similarity matrix of the molecules to be tested based on the matching order comprises: Based on the matching order, determining the sequence cluster association group of the molecule to be tested; Configuring an association group similarity matching weight of the sequence cluster association group; Calculate the similarity of the matching items of the sequence cluster association group corresponding to the similar matching items by using the molecular structure fingerprint corresponding to the molecule to be tested; Configure the item similarity matching weight of the similarity matching item corresponding to the sequence cluster association group; The similarity matrix of the molecule to be tested is constructed by using the association group similarity matching weights, matching item similarities and item similarity matching weights.

10. A molecular similarity calculation device based on molecular structure fingerprint, characterized in that: The device comprises: A multi-scale model building module, used to obtain structural information of the molecule to be tested, build a multi-scale model of the molecule to be tested based on the structural information, and determine the atomic type, chemical bond type, and spatial conformation of the molecule to be tested based on the multi-scale model; A molecular structure fingerprint construction module is used to construct a molecular graph of the molecule to be tested based on the atomic type, chemical bond type, and spatial conformation, determine the molecular characteristics of the molecule to be tested based on the molecular graph and a preset structural GNN model, determine the molecular fingerprint category of the molecule to be tested in a preset molecular fingerprint library, and construct a molecular structure fingerprint of the molecule to be tested based on the molecular characteristics and the molecular fingerprint category; A cluster association group construction module is used to analyze the structural parameters of the molecule to be tested based on the molecular structure fingerprint, construct similar matching items of the molecule to be tested through the structural parameters, calculate the correlation coefficient of the similar matching items, and construct the cluster association group of the molecule to be tested based on the correlation coefficient and the similar matching items; A maximum similarity matching item analysis module, used to determine the maximum similarity matching item of the cluster association group and calculate the similarity measure of the maximum similarity matching item; The molecular similarity calculation module is used to determine the matching order of the cluster association groups through the similarity metric, construct a similarity matrix of the molecules to be tested based on the matching order, and analyze the similarity of the molecules to be tested based on the similarity matrix.

Images