Molecular similarity calculation method and apparatus based on photonic quantum computer

Abstract

The present invention relates to the technical field of drug design, and in particular to a molecular similarity calculation method based on a photonic quantum computer, comprising: on the basis of molecular structures, respectively converting two molecules requiring molecular similarity calculation into molecular graphs, wherein vertexes of each molecular graph represent atoms in the corresponding molecule, and edges of the molecular graph represent chemical bond connections between the atoms in the molecule; using the two molecular graphs to construct a conflict graph, wherein vertexes of the conflict graph are derived from pairwise vertex matching mappings between the two molecular graphs, and edges of the conflict graph are derived from discrimination results between the vertices in the conflict graph; solving a maximum weighted independent subset of the conflict graph by means of a photonic quantum computer and obtaining information of each vertex of the conflict graph in the maximum weighted independent subset; and calculating the similarity between the two molecules on the basis of the information of each vertex of the conflict graph in the maximum weighted independent subset. The method overcomes the defects in conventional solving approaches of low solving speed and high false positive rate, effectively improving the calculation speed and accuracy.

Description

Molecular similarity calculation method and device based on optical quantum computer Technical Field

[0001] The present invention belongs to the technical field of drug design, and relates to a molecular similarity calculation method and device, and in particular to a molecular similarity calculation method and device based on an optical quantum computer. Background Art

[0002] In the field of drug design, drug development is costly, time-consuming, and has a high failure rate. Therefore, researchers are constantly searching for more cost-effective and successful drug development methods. Traditional drug discovery can be categorized into receptor-based and ligand-based drug screening. Ligand-based drug screening primarily identifies molecules similar to known active molecular groups as lead compounds, which are then further screened and validated as drug candidates.

[0003] The existing method is to compare the similarity of different molecules through molecular fingerprints. Molecular fingerprints are created by encoding molecular structural fragments into characteristic binary vectors, where each bit corresponds to the presence of a specific fragment. If two molecular fingerprints have a 1 at the same position, then both molecules have the same fragment, and the more fragments they share, the more similar they are. This method compares the similarity of molecules at the 2D level, but does not use all the effective information contained in the molecules, such as the 3D conformation of the molecules. The 3D conformation of the molecule has an important influence on protein binding in real reaction systems, and the molecular similarity obtained based on the molecular fingerprint method will lose this part of information. Therefore, the similarity calculation of molecular fingerprints will lead to a high false positive rate, that is, the high-similarity compounds screened out by calculation are not active in the real experimental system, and screening in a compound library with a large number of molecules is time-consuming.

[0004] Therefore, in view of the defects in the above-mentioned prior art, it is necessary to develop a new molecular similarity calculation method and device. Summary of the Invention

[0005] In order to overcome the defects of the existing technology, the present invention proposes a molecular similarity calculation method and device based on an optical quantum computer, which solves the shortcomings of slow solution speed and high false positive rate of traditional solution methods, and effectively improves the calculation speed and accuracy.

[0006] In order to achieve the above object, the present invention provides the following technical solutions:

[0007] A method for calculating molecular similarity based on an optical quantum computer, characterized by comprising the following steps:

[0008] Based on the molecular structure, the two molecules whose molecular similarity needs to be calculated are converted into molecular graphs respectively, wherein the vertices of the molecular graph represent the atoms in the molecules, and the edges of the molecular graph represent the chemical bonds between the atoms in the molecules;

[0009] Constructing a conflict graph using the two molecular graphs, wherein the vertices of the conflict graph are derived from the matching mapping between each two vertices in the two molecular graphs, and the edges of the conflict graph are derived from the discrimination results between the vertices in the conflict graphs;

[0010] Solving the maximum weighted independent subset of the conflict graph by an optical quantum computer and obtaining information of each vertex of the conflict graph in the maximum weighted independent subset;

[0011] The similarity between the two molecules is calculated by using the information of each vertex of the conflict graph in the maximum weighted independent subset.

[0012] Preferably, the vertices of the molecular graph contain characteristic information of atoms and the edges of the molecular graph contain chemical bond connection information between atoms, the characteristic information of atoms includes the atom name, the number of visible hydrogens, the number of implicit hydrogens, the number of adjacent atoms, the explicit valence, the formal charge and whether it is an intra-ring atom, and the chemical bond connection information between atoms includes the type of bond, the aromaticity of the bond, whether it is an intra-ring bond and the stereo information of the bond.

[0013] Preferably, when constructing a conflict graph using two molecular graphs, matched atom pairs are obtained by comparing the similarities between the atoms in the original molecules corresponding to the two molecular graphs and the matched atom pairs are used as vertices of the conflict graph. Whether the difference in distances between the two atoms in the original molecules corresponding to the vertices in the two molecular graphs contained in the two vertices in the conflict graph is greater than a threshold is compared to determine whether the two vertex mappings in the conflict graph conflict, thereby determining whether to add an edge between the two vertices.

[0014] Preferably, obtaining matching atom pairs by comparing the similarities between atoms in the original molecules corresponding to the two molecular graphs and using the matching atom pairs as vertices of the conflict graph is specifically as follows:

[0015] Calculate the similarity between the atoms of the two original molecules corresponding to the molecular graphs, wherein the similarity between the two atoms is calculated as follows:

[0016] Where N represents the total number of atomic characteristic information, n represents the ordinal number of atomic characteristic information, Gf n and G'f nThey represent the value of the nth characteristic information of the atoms in the original molecule corresponding to the first molecular graph and the value of the nth characteristic information of the atoms in the original molecule corresponding to the second molecular graph, min(Gf n ,G'f n ) represents the smaller value of the nth characteristic information of the atoms in the original molecule corresponding to the first molecular graph and the nth characteristic information of the atoms in the original molecule corresponding to the second molecular graph, max(Gf n ,G'f n ) represents the larger value of the nth characteristic information of the atoms in the original molecule corresponding to the first molecular graph and the nth characteristic information of the atoms in the original molecule corresponding to the second molecular graph;

[0017] Comparing the similarity between the two atoms with a threshold S, and when the similarity is greater than the threshold S, it is considered that there is a mapping between the atom in the original molecule corresponding to the first molecular graph and the atom in the original molecule corresponding to the second molecular graph, and the two atoms are a matching atom pair, and the matching atom pair is used as a vertex of the conflict graph;

[0018] Calculate the pairwise atomic similarity of all atoms in the original molecule corresponding to the first molecular graph and all atoms in the original molecule corresponding to the second molecular graph, find all matching atom pairs, and obtain the vertex set with vertex weights of the conflict graph. Where, v i Indicates the i-th atom in the original molecule corresponding to the first molecular graph, v' j represents the jth atom in the original molecule corresponding to the second molecular graph, Represents the vertex weight of the vertex in the conflict graph.

[0019] Preferably, the threshold S is 0.75.

[0020] Preferably, whether two vertex mappings in the conflict graph conflict is determined by comparing whether the difference in distances between two atoms in the original molecule corresponding to the vertices in the two molecular graphs contained in the two vertices in the conflict graph is greater than a threshold, thereby determining whether to add an edge between the two vertices. Specifically,

[0021] Compare whether the distance difference between the two atoms in the original molecule corresponding to the two vertices in the molecular graph contained in the two vertices in the conflict graph is greater than the threshold dt. If it is greater than the threshold dt, then add an edge to the two vertices. The specific mathematical expression of the judgment is as follows: E c ={(v i ,v' j ),(v k ,v' l )||d(v i ,vk )-d(v' j ,v' l )|>d t}

[0022] Where, E c represents the edge set between the vertices in the conflict graph, (v i ,v' j ) and (v k ,v' l ) represent two vertices in the conflict graph, d(v i ,v k ) and d(v' j ,v' l ) represent the vertices v in the first molecular graph. i and v k The geometric distance between the two atoms in the corresponding original molecule and the vertex v' in the second molecular graph j and v' l The geometric distance between the two atoms in the corresponding original molecule.

[0023] Preferably, the threshold dt is 2, and an edge weight is assigned to the edge of the conflict graph, where the value of the edge weight is the smaller value of the vertex weights of the two connected vertices plus 1.

[0024] Preferably, when solving the maximum weighted independent subset of the conflict graph, the following QUBO model is constructed:

[0025] Where H represents the Hamiltonian, which is the target value for optimization; p and q represent the p-th vertex and the q-th vertex in the conflict graph, respectively; x p and x q are two binary variables, which are equal to 1 if the p-th vertex and the q-th vertex of the conflict graph are respectively included in the maximum weighted independent subset, and 0 otherwise; p is the vertex weight of the p-th vertex in the conflict graph; w pq V represents the edge weight of the edge connecting the p-th vertex and the q-th vertex in the conflict graph; c and E c Represent the vertex set and edge set in the conflict graph respectively.

[0026] Preferably, the similarity between the two molecules is calculated by using the information of each vertex of the conflict graph in the maximum weighted independent subset. The specific calculation method is as follows:

[0027] In the formula, G and G' represent the two molecules whose molecular similarity needs to be calculated, S(G,G') represents the similarity between molecules G and G', |V1| and |V2| represent the number of atoms in molecules G and G' respectively, and They respectively represent the number of atoms belonging only to molecule G and only to molecule G' in each vertex of the conflict graph in the maximum weighted independent subset, and δ is an adjustable parameter.

[0028] In addition, the present invention also provides a molecular similarity calculation device based on an optical quantum computer, which is characterized by comprising:

[0029] one or more processors;

[0030] a memory for storing one or more programs;

[0031] When the one or more programs are executed by the one or more processors, the one or more processors are enabled to implement the molecular similarity calculation method based on an optical quantum computer according to any one of claims 1 to 9.

[0032] Compared with the prior art, the molecular similarity calculation method and device based on an optical quantum computer of the present invention have one or more of the following beneficial technical effects:

[0033] 1. This invention solves the problem of calculating molecular similarity by designing a solution model suitable for optical quantum computers, overcoming the shortcomings of traditional solution methods such as slow solution speed and high false positive rate, and effectively improving the calculation speed and accuracy.

[0034] 2. This invention improves computing efficiency and solution results by designing a mathematical model suitable for quantum computing and then using an optical quantum computer to accelerate the solution of the problem, thus solving the problem that the existing technology is difficult to calculate and cannot obtain accurate solutions for complex scheduling problems.

[0035] 3. Traditional problem-solving methods can only obtain approximate solutions to the problem rather than the global optimal solution, and therefore cannot obtain a result with relatively high accuracy. The present invention uses QUBO (Quadratic Unconstrained Binary Optimization) modeling and utilizes an optical quantum computer to obtain the optimal solution to the problem. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] FIG1 is a flow chart of the molecular similarity calculation method based on an optical quantum computer of the present invention.

[0037] Figure 2 is the ROC curve of COMT molecules solved based on optical quantum computers. DETAILED DESCRIPTION

[0038] Before describing in detail any embodiment of the present invention, it should be understood that the present invention is not limited in its application to the construction and arrangement details of the components set forth in the following description or illustrated in the following figures. The present invention is capable of other embodiments and can be practiced or carried out in various ways. In addition, it should be understood that the words and terms used herein are for descriptive purposes and should not be considered restrictive. The use of "including" or "having" and their variations herein is intended to cover the items and their equivalents set forth below and additional items. Unless otherwise specified or limited, the terms "mounted", "connected", "supported" and "coupled" and their variations are used broadly and cover direct mounting and indirect mounting, connection, support and coupling. In addition, "connected" and "coupled" are not limited to physical or mechanical connections or couplings.

[0039] Furthermore, in the disclosure of the present invention, the term "one" should be understood as "at least one" or "one or more", that is, in one embodiment, the number of an element may be one, while in another embodiment, the number of the elements may be multiple, and the term "one" should not be understood as a limitation on the quantity.

[0040] Molecular similarity is often used to screen molecules with similar structures to known active groups. The traditional method for calculating molecular similarity is to first generate molecular fingerprints for the two molecules and then obtain similarity information by comparing the molecular fingerprints of the two molecules. However, the similarity calculation method based on molecular fingerprints does not use all the effective information contained in the molecules, such as the 3D conformation of the molecules. The 3D conformation of the molecule has a significant impact on protein binding in real reaction systems, and the molecular similarity obtained based on the molecular fingerprint method loses this part of information. Therefore, similarity calculations based on molecular fingerprints will result in a high false positive rate. In other words, the highly similar compounds screened by calculation are not active in the real experimental system, and screening in a compound library with a large number of molecules is time-consuming.

[0041] To overcome the above-mentioned drawbacks, the present invention utilizes the 3D information of molecules and first converts a single molecule into a graph structure, wherein the vertices of the molecular graph represent the atoms in the original molecule, and the edges of the molecular graph represent the chemical bond connections between the atoms in the original molecule. In order to solve the similarity of two molecules, a conflict graph is first constructed using the two molecular graphs. The vertices of the conflict graph are derived from the matching mapping between the two vertices in the two molecular graphs. If the matching requirements are met, a vertex is generated. The connection of the edges in the conflict graph is derived from the discrimination results between the vertices in the conflict graph. If the difference in the geometric distance between the two vertices in the original molecular graphs contained between the two vertices in the conflict graph is greater than a threshold, an edge is added to the two vertices in the conflict graph, and a weight is assigned to the edge, which is the minimum value of the weight connecting the two vertices plus 1. By constructing a conflict graph, the problem of solving molecular similarity is first converted into solving the maximum weighted independent subset of the conflict graph, and then a combinatorial optimization mathematical model suitable for quantum computing is designed to solve the problem. Finally, an optical quantum computer is used to solve the optimal solution to the problem, obtain the maximum weighted independent subset and obtain the information of each vertex in the set, and calculate the molecular similarity based on the vertex information.

[0042] FIG1 shows a flow chart of the molecular similarity calculation method based on an optical quantum computer of the present invention. As shown in FIG1 , the molecular similarity calculation method based on an optical quantum computer of the present invention comprises the following steps:

[0043] 1. Based on the molecular structure, the two molecules whose molecular similarity needs to be calculated are converted into molecular graphs respectively.

[0044] The vertices of the molecular graph represent the atoms in the molecule, and the edges of the molecular graph represent the chemical bonds between the atoms in the molecule.

[0045] For the convenience of information extraction and calculation, the present invention first converts a single molecule into a structure represented by a molecular graph based on the molecular structure. The specific conversion process is: the vertices in the molecular graph are derived from the various atoms in the original molecule, and the vertices contain the characteristic information of the atoms. These characteristic information are extracted according to the smiles format file of the molecule using RDKit (RDKit is an open source toolkit for chemical informatics), mainly including: the name of the atom, the number of displayed hydrogens, the number of implicit hydrogens, the number of adjacent atoms, the explicit valence, the formal charge, and whether it is an intra-ring atom. The connection of the edges between the vertices in the molecular graph is derived from the chemical bond connection between the various atoms in the original molecule. The edges in the molecular graph contain the chemical bond connection information in the original molecule, mainly including the type of bond (single bond, double bond, triple bond, conjugated), the aromaticity of the bond, whether it is an intra-ring bond, the three-dimensional information of the bond, etc.

[0046] Based on the above process, the molecular graph of the two molecules whose molecular similarity needs to be calculated is obtained for the next calculation.

[0047] 2. Construct a conflict graph using the two molecular graphs.

[0048] In the first step, molecular graphs of two molecules whose molecular similarity needs to be calculated are obtained, and then a conflict graph is constructed based on the two molecular graphs.

[0049] In the present invention, the vertices of the conflict graph are derived from the matching mapping between the two vertices in the two molecular graphs. That is, by comparing the similarities between the atoms in the original molecules corresponding to the two molecular graphs, matching atom pairs are obtained and used as the vertices of the conflict graph.

[0050] Specifically, first, the similarity between the atoms of the original molecules corresponding to the two molecular graphs is calculated. The similarity between the two atoms is calculated as follows:

[0051] Wherein, N represents the total number of characteristic information of the atom (as mentioned above, in the present invention, there are seven characteristic information of the atom, namely, the atom name, the number of displayed hydrogens, the number of implicit hydrogens, the number of adjacent atoms, the explicit valence, the formal charge, and whether it is an intra-ring atom, so N = 7), n represents the ordinal number of the characteristic information of the atom (in the present invention, n is a value from 1 to 7), Gf n and G'f n They represent the value of the nth characteristic information of the atoms in the original molecule corresponding to the first molecular graph and the value of the nth characteristic information of the atoms in the original molecule corresponding to the second molecular graph, min(Gf n ,G'f n ) represents the smaller value of the nth characteristic information of the atoms in the original molecule corresponding to the first molecular graph and the nth characteristic information of the atoms in the original molecule corresponding to the second molecular graph, max(Gf n ,G'f n ) represents the larger value between the value of the nth characteristic information of the atoms in the original molecule corresponding to the first molecular graph and the value of the nth characteristic information of the atoms in the original molecule corresponding to the second molecular graph.

[0052] Next, the similarity between the two atoms is compared with a threshold S. If the similarity is greater than the threshold S, it is considered that there is a mapping between the atom in the original molecule corresponding to the first molecular graph and the atom in the original molecule corresponding to the second molecular graph. The two atoms are then considered a matching atom pair, and the matching atom pair is used as a vertex in the conflict graph. In the present invention, the threshold S is 0.75.

[0053] Next, all atoms in the original molecule corresponding to the first molecular graph and all atoms in the original molecule corresponding to the second molecular graph are subjected to pairwise atomic similarity calculations to find all matching atom pairs and obtain the vertex set with vertex weights of the conflict graph. Where, v i Indicates the i-th atom in the original molecule corresponding to the first molecular graph, v' j represents the jth atom in the original molecule corresponding to the second molecular graph, The vertex weight of the vertex in the conflict graph is represented. Each vertex in the vertex set represents a possible atomic pairing in the original two molecules, and its vertex weight shows its contribution to the similarity of the entire molecule.

[0054] Next, consider the edge connections in the conflict graph. The edges of the conflict graph are derived from the judgment results between the vertices in the conflict graph. Specifically, by comparing whether the difference in distance between the two atoms in the original molecule corresponding to the vertices in the molecular graphs contained in the two vertices in the conflict graph is greater than a threshold, we determine whether the two vertex mappings in the conflict graph conflict, thereby determining whether to add an edge between the two vertices.

[0055] Specifically, the distance difference between the two atoms in the original molecule corresponding to the two vertices in the molecular graph contained in the two vertices in the conflict graph is compared to see whether it is greater than the threshold dt. If it is greater than the threshold dt, it means that the two vertex mappings in the conflict graph are conflicting, then edges are added to the two vertices. The specific mathematical expression of the judgment is as follows: E c ={(v i ,v' j ),(v k ,v' l )||d(v i ,v k )-d(v' j ,v' l )|>d t}

[0056] Among them, E c represents the edge set between the vertices in the conflict graph, (v i ,v' j ) and (v k ,v' l ) represent two vertices in the conflict graph, d(v i ,v k ) and d(v' j ,v' l ) represent the vertices v in the first molecular graph. i and v k The geometric distance between the two atoms in the corresponding original molecule and the vertex v' in the second molecular graphj and v' l The geometric distance between the two atoms in the corresponding original molecule.

[0057] In the present invention, when adding an edge to two vertices, an edge weight is assigned to the edge of the conflict graph, and the value of the edge weight is the smaller value of the vertex weights of the two connected vertices plus 1. In addition, the threshold dt is a variable threshold parameter, which is 2 in the present invention.

[0058] Through the above steps, a conflict graph with vertices and edges each having weights is obtained.

[0059] 3. Solve the maximum weighted independent subset of the conflict graph through an optical quantum computer and obtain information of each vertex of the conflict graph in the maximum weighted independent subset.

[0060] After constructing the conflict graph, the molecular similarity problem can be converted into solving the maximum weighted independent subset problem based on the conflict graph. The vertices in the conflict graph represent the possible pairwise vertex matching mappings in the original two molecules, and the edges connecting the vertices of the conflict graph represent the conflict relationships between the vertices of the conflict graph. The vertices and edges of the conflict graph are weighted. Therefore, the solution goal is to obtain a set of independent vertices that maximizes the total weight. The maximum weighted independent subset problem is a classic NP-hard problem, and it is difficult to obtain the optimal solution using a classical computer. Establishing a QUBO model for this problem and solving it using a quantum computer is an effective solution to the NP-hard problem. The following QUBO model can be constructed for solving the maximum weighted independent subset of the conflict graph in this problem:

[0061] Where H represents the Hamiltonian, which is the target value of optimization; p and q represent the p-th vertex and the q-th vertex in the conflict graph respectively; x p and x q are two binary variables, which are equal to 1 if the p-th vertex and the q-th vertex of the conflict graph are respectively included in the maximum weighted independent subset, and 0 otherwise; p is the vertex weight of the p-th vertex in the conflict graph; w pq V represents the edge weight of the edge connecting the p-th vertex and the q-th vertex in the conflict graph; c and E c For example, the maximum number of vertices and edges in the conflict graph of two molecules containing 20 atoms each is 400, and the maximum number of edges is 160,000.

[0062] Then, according to the above QUBO model, an optical quantum computer is used to solve the problem. The QUBO model is applicable to special quantum computers such as quantum annealing machines and coherent Ising machines (CIM). The implementation and solution of the physical machine takes the CIM based on the degenerate optical parametric oscillator (DOPO) as an example. This is a hybrid quantum computing system consisting of an optical part and an electrical part. The optical part includes a laser, an amplifier, a periodically poled lithium niobate crystal (PPLN) and an optical fiber loop. The laser uses a femtosecond pulse fiber laser and is equipped with an amplifier system. The amplified laser is first frequency-doubled using a PPLN crystal. The frequency-doubled laser is used as a pump source to synchronously pump the PPLN crystal in a fiber loop to form a degenerate optical parametric oscillation. Hundreds of oscillation pulses can exist simultaneously in the fiber loop. The electrical system includes an FPGA (Field Programmable Gate Array), AD / DA (digital-to-analog / analog-to-digital converter), and a phase detector. The phase detector measures the laser output from the fiber loop and the fundamental frequency laser, allowing for phase measurement. The FPGA, combined with the high-speed AD / DA, enables optical pulse measurement and feedback control.

[0063] Unlike classical computers, which operate on semiconductor integrated circuits, CIM uses laser pulses in optical fibers as quantum bits for computation. In DOPO, pump light is incident on a nonlinear optical crystal, splitting it into two beams. These two beams have the same polarization direction and a frequency half that of the pump light, remaining in a compressed state and functioning as a quantum bit. By gradually increasing the pump light power above the oscillation threshold, the generated light becomes coherent, with its phase splitting into two states (phase 0 and π). This phase can then be set to ±1 relative to the spin, enabling optimization problems.

[0064] Using an optical quantum computer, a solution is found that minimizes the target value H. This solution is a set of 0, 1 variables, such as [0, 1, 1, 1, 0, 0, 1, 0, ...]. The number of elements in this list is the number of vertices in the conflict graph. A 1 in the solution indicates that the corresponding vertex is included in the maximum weighted independent subset of the conflict graph, while a 0 indicates that it is not. This maximum weighted independent subset can be used to obtain vertex information from both molecular graphs.

[0065] Fourth, the similarity between the two molecules is calculated using the information of each vertex of the conflict graph in the maximum weighted independent subset.

[0066] After finding the maximum weighted independent subset, the vertex information of the conflict graph contained in the maximum weighted independent subset can be obtained, and the similarity of the two molecules can be calculated based on the vertex information. The calculation method is as follows:

[0067] In the formula, G and G' represent the two molecules whose molecular similarity needs to be calculated, S(G,G') represents the similarity between molecules G and G', |V1| and |V2| represent the number of atoms in molecules G and G' respectively, and They respectively represent the number of atoms belonging only to molecule G and only to molecule G' in each vertex of the conflict graph in the maximum weighted independent subset, and δ is an adjustable parameter.

[0068] For example, there are 10 and 12 vertices (i.e., 10 and 12 atoms) in molecules G and G' respectively. The vertex set in the obtained maximum weighted independent subset contains 5 unique atoms in molecule G and 4 unique atoms in molecule G'. Then,

[0069] Where δ is an adjustable parameter, which is set to 0.5 in this example.

[0070] The present invention solves the calculation problem of molecular similarity by designing a solution model suitable for optical quantum computers, overcoming the shortcomings of traditional solution methods such as slow solution speed and high false positive rate, and effectively improving the calculation speed and accuracy.

[0071] In order to verify the accuracy of the model for active molecule screening, the present invention utilizes the model in conjunction with optical quantum computer solution, and is verified on the data set DUD_LIB_VS_1.0 library. The data set is a data set that is verified and used to evaluate the performance of the screening method, comprising multiple protein classes, each target protein class containing active molecules (activities) set and inactive molecules (decoys) set, active molecules are molecules that can be combined with the target protein through experimental verification, and inactive molecules are molecules with similar structures but no relevant binding activity. In addition, a query molecule is also included, which is the ligand molecule included when parsing the target protein structure, and has the optimal conformation combined with the target protein. By calculating the similarity of the query molecule to the active molecule set and inactive molecules, and screening according to a threshold value, the accuracy of the final calculation screening and the screening effect of the head molecule are evaluated to evaluate the computational performance of the model.

[0072] The present invention selects catechol-o-methyltransferase (COMT) proteins as an example to compare the effect of solving based on optical quantum computers (that is, the effect of the present invention) with the traditional molecular fingerprint-based similarity calculation method (MFP) and the graph-based similarity calculation method (GMS). COMT is a dopamine-degrading enzyme and one of the main regulatory pathways of dopamine in the brain. It has been confirmed to be a drug biological target for treating various central and peripheral nervous system diseases, including Parkinson's disease, depression, schizophrenia and other dopamine-deficiency-related diseases. The present invention calculated the similarity of 8 active molecules and 144 inactive molecules of the COMT class, and the calculation results are shown in Table 1.

[0073] Table 1 Calculation results of molecular similarity of COMT class

[0074] In the biomedical field, the receiver operating characteristic (ROC) curve is often used to evaluate the quality of a screening model. This curve sets multiple critical values ​​for a continuous variable, calculates the corresponding sensitivity and specificity at each critical value, and then plots the curve with sensitivity as the vertical axis and 1-specificity as the horizontal axis. The area under the ROC curve (AUC) can be used as a quantitative indicator to judge the quality of the model's classification. The ROC curve of the molecular similarity model solved by the optical quantum computer for the COMT class is shown in Figure 2, with an AUC of 98%. Therefore, the model achieved a very ideal effect for screening the entire data set.

[0075] In the pharmaceutical field, virtual screening of large compound libraries can only test the bioactivity of a small subset of compounds. Therefore, in addition to the classification performance of the entire dataset, the early enrichment capability of the model in virtual screening may be of greater concern. ROCE0.5% and ROCE1% represent the ratio of the true positive rate to the false positive rate at false positive rates of 0.5% and 1%, respectively. A higher value indicates better early enrichment capability. Calculations of the COMT class using an optical quantum computer yielded ROCE0.5% and ROCE1% of 125 and 62.5, respectively, surpassing those of the traditional MFP method and the graph-based similarity calculation method GMS (Table 2).

[0076] Table 2 Comparison of the results of the solution based on optical quantum computers with those of the GMS and MFP methods

[0077] Finally, the present invention also relates to a molecular similarity calculation device based on an optical quantum computer, which includes: one or more processors; a memory for storing one or more programs; when the one or more programs are executed by the one or more processors, the one or more processors implement the molecular similarity calculation method based on the optical quantum computer as described above.

[0078] Finally, it should be noted that the above embodiments are intended only to illustrate the technical solutions of the present invention and are not intended to limit the scope of protection of the present invention. Those skilled in the art may, based on the principles of the present invention, modify or replace the technical solutions of the present invention with equivalents without departing from the essence and scope of the technical solutions of the present invention.

Claims

1. A molecular similarity calculation method based on an optical quantum computer, characterized in that: The following steps are involved: Based on the molecular structure, the two molecules whose molecular similarity needs to be calculated are converted into molecular graphs respectively, wherein the vertices of the molecular graph represent the atoms in the molecules, and the edges of the molecular graph represent the chemical bonds between the atoms in the molecules; Constructing a conflict graph using the two molecular graphs, wherein the vertices of the conflict graph are derived from the matching mapping between each two vertices in the two molecular graphs, and the edges of the conflict graph are derived from the discrimination results between the vertices in the conflict graphs; Solving the maximum weighted independent subset of the conflict graph by an optical quantum computer and obtaining information of each vertex of the conflict graph in the maximum weighted independent subset; The similarity between the two molecules is calculated by using the information of each vertex of the conflict graph in the maximum weighted independent subset.

2. The molecular similarity calculation method based on optical quantum computer according to claim 1, characterized in that: The vertices of the molecular graph contain characteristic information of atoms and the edges of the molecular graph contain chemical bond connection information between atoms. The characteristic information of atoms includes the atom name, the number of visible hydrogens, the number of implicit hydrogens, the number of adjacent atoms, the explicit valence, the formal charge and whether it is an intra-ring atom. The chemical bond connection information between atoms includes the bond type, the aromaticity of the bond, whether it is an intra-ring bond and the stereo information of the bond.

3. The molecular similarity calculation method based on optical quantum computer according to claim 1, characterized in that: When constructing a conflict graph using the two molecular graphs, matching atom pairs are obtained by comparing the similarities between the atoms in the original molecules corresponding to the two molecular graphs and the matching atom pairs are used as vertices of the conflict graph. Whether the two vertex mappings in the conflict graph conflict is determined by comparing whether the difference in distances between the two atoms in the original molecules corresponding to the vertices in the two molecular graphs contained in the two vertices in the conflict graph is greater than a threshold, thereby determining whether to add an edge between the two vertices.

4. The molecular similarity calculation method based on optical quantum computer according to claim 3, characterized in that: The method of obtaining matching atom pairs by comparing the similarities between atoms of the original molecules corresponding to the two molecular graphs and using the matching atom pairs as vertices of the conflict graph is as follows: Calculate the similarity between the atoms of the two original molecules corresponding to the molecular graphs, wherein the similarity between the two atoms is calculated as follows: Where N represents the total number of atomic characteristic information, n represents the ordinal number of atomic characteristic information, Gf n and G'f n They represent the value of the nth characteristic information of the atoms in the original molecule corresponding to the first molecular graph and the value of the nth characteristic information of the atoms in the original molecule corresponding to the second molecular graph, min(Gf n ,G'f n ) represents the smaller value of the nth characteristic information of the atoms in the original molecule corresponding to the first molecular graph and the nth characteristic information of the atoms in the original molecule corresponding to the second molecular graph, max(Gf n ,G'f n ) represents the larger value of the nth characteristic information of the atoms in the original molecule corresponding to the first molecular graph and the nth characteristic information of the atoms in the original molecule corresponding to the second molecular graph; Comparing the similarity between the two atoms with a threshold S, and when the similarity is greater than the threshold S, it is considered that there is a mapping between the atom in the original molecule corresponding to the first molecular graph and the atom in the original molecule corresponding to the second molecular graph, and the two atoms are a matching atom pair, and the matching atom pair is used as a vertex of the conflict graph; Calculate the pairwise atomic similarity of all atoms in the original molecule corresponding to the first molecular graph and all atoms in the original molecule corresponding to the second molecular graph, find all matching atom pairs, and obtain the vertex set with vertex weights of the conflict graph. Where, v i Indicates the i-th atom in the original molecule corresponding to the first molecular graph, v' j represents the jth atom in the original molecule corresponding to the second molecular graph, Represents the vertex weight of the vertex in the conflict graph.

5. The molecular similarity calculation method based on optical quantum computer according to claim 4, characterized in that: The threshold S is 0.

75.

6. The molecular similarity calculation method based on optical quantum computer according to claim 5, characterized in that: Whether the two vertex mappings in the conflict graph conflict is determined by comparing whether the difference in distance between the two atoms in the original molecule corresponding to the vertices in the two molecular graphs contained in the two vertices in the conflict graph is greater than a threshold, thereby determining whether to add an edge between the two vertices. Specifically, Compare whether the distance difference between the two atoms in the original molecule corresponding to the two vertices in the molecular graph contained in the two vertices in the conflict graph is greater than the threshold dt. If it is greater than the threshold dt, then add an edge to the two vertices. The specific mathematical expression of the judgment is as follows: E c ={(v i ,v' j ),(v k ,v' l )||d(v i ,v k )-d(v' j ,v' l )|>d t } Where, E c represents the edge set between the vertices in the conflict graph, (v i ,v' j ) and (v k ,v' l ) represent two vertices in the conflict graph, d(v i ,v k ) and d(v' j ,v' l ) represent the vertices v in the first molecular graph. i and v k The geometric distance between the two atoms in the corresponding original molecule and the vertex v' in the second molecular graph j and v' l The geometric distance between the two atoms in the corresponding original molecule.

7. The molecular similarity calculation method based on optical quantum computer according to claim 6, characterized in that: The threshold dt is 2, and an edge weight is assigned to the edge of the conflict graph, where the value of the edge weight is the smaller value of the vertex weights of the two connected vertices plus 1.

8. The molecular similarity calculation method based on optical quantum computer according to claim 7, characterized in that: When solving the maximum weighted independent subset of the conflict graph, the following QUBO model is constructed: Where H represents the Hamiltonian, which is the target value for optimization; p and q represent the p-th vertex and the q-th vertex in the conflict graph, respectively; x p and x q are two binary variables, which are equal to 1 if the p-th vertex and the q-th vertex of the conflict graph are respectively included in the maximum weighted independent subset, and 0 otherwise; p is the vertex weight of the p-th vertex in the conflict graph; w pq V represents the edge weight of the edge connecting the p-th vertex and the q-th vertex in the conflict graph; c and E c Represent the vertex set and edge set in the conflict graph respectively.

9. The molecular similarity calculation method based on optical quantum computer according to claim 8, characterized in that: The specific calculation method of calculating the similarity of the two molecules by using the information of each vertex of the conflict graph in the maximum weighted independent subset is as follows: Among them, G and G' represent the two molecules whose molecular similarity needs to be calculated, S(G,G') represents the similarity between molecules G and G', |V1| and |V2| represent the number of atoms in molecules G and G' respectively, and They respectively represent the number of atoms belonging only to molecule G and only to molecule G' in each vertex of the conflict graph in the maximum weighted independent subset, and δ is an adjustable parameter.

10. A molecular similarity calculation device based on an optical quantum computer, characterized in that: include: one or more processors; a memory for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors are enabled to implement the molecular similarity calculation method based on an optical quantum computer according to any one of claims 1 to 9.

Images