CN114446383B - Quantum calculation-based ligand-protein interaction prediction method - Google Patents
Quantum calculation-based ligand-protein interaction prediction method Download PDFInfo
- Publication number
- CN114446383B CN114446383B CN202210079577.0A CN202210079577A CN114446383B CN 114446383 B CN114446383 B CN 114446383B CN 202210079577 A CN202210079577 A CN 202210079577A CN 114446383 B CN114446383 B CN 114446383B
- Authority
- CN
- China
- Prior art keywords
- protein
- ligand
- molecule
- quantum
- docking
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
- G16B5/00—ICT specially adapted for modelling or simulations in systems biology, e.g. gene-regulatory networks, protein interaction networks or metabolic networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N10/00—Quantum computing, i.e. information processing based on quantum-mechanical phenomena
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
- G16B20/00—ICT specially adapted for functional genomics or proteomics, e.g. genotype-phenotype associations
- G16B20/30—Detection of binding sites or motifs
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
- G16B40/00—ICT specially adapted for biostatistics; ICT specially adapted for bioinformatics-related machine learning or data mining, e.g. knowledge discovery or pattern finding
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
- G16C20/00—Chemoinformatics, i.e. ICT specially adapted for the handling of physicochemical or structural data of chemical particles, elements, compounds or mixtures
- G16C20/50—Molecular design, e.g. of drugs
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
- G16C20/00—Chemoinformatics, i.e. ICT specially adapted for the handling of physicochemical or structural data of chemical particles, elements, compounds or mixtures
- G16C20/70—Machine learning, data mining or chemometrics
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Theoretical Computer Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Medical Informatics (AREA)
- General Health & Medical Sciences (AREA)
- Chemical & Material Sciences (AREA)
- Spectroscopy & Molecular Physics (AREA)
- Software Systems (AREA)
- Biotechnology (AREA)
- Evolutionary Biology (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Biophysics (AREA)
- Computing Systems (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Molecular Biology (AREA)
- General Physics & Mathematics (AREA)
- Databases & Information Systems (AREA)
- Crystallography & Structural Chemistry (AREA)
- Analytical Chemistry (AREA)
- Condensed Matter Physics & Semiconductors (AREA)
- Genetics & Genomics (AREA)
- Medicinal Chemistry (AREA)
- Pharmacology & Pharmacy (AREA)
- Epidemiology (AREA)
- Bioethics (AREA)
- Physiology (AREA)
- Computational Mathematics (AREA)
- Public Health (AREA)
- Proteomics, Peptides & Aminoacids (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
Abstract
The invention discloses a method for predicting ligand-protein interaction based on quantum computation, which belongs to the field of quantum computation and computation biology and comprises the following steps: obtaining a 3D structure of the ligand drug molecule and the protein molecule; predicting the 3D structure of the protein molecule by using a quantum hidden Markov model to obtain a possible binding site on the protein molecule; docking a ligand drug molecule with the binding site and extracting features affecting ligand-protein docking using a quantum convolutional neural network; scoring the docking process according to characteristics affecting ligand-protein docking, and predicting the effect of ligand-protein interaction. According to the invention, based on the related information of molecular docking description of a quantum convolution neural network, a quantum hidden Markov model is used for predicting possible binding sites on receptor proteins, so that synchronous prediction of molecular docking and protein structure is realized, and meanwhile, quantum calculation is combined with the field of biological medicine, so that the difficulty in predicting the 3D structure of the protein is solved.
Description
Technical Field
The invention relates to the field of quantum computing and computing biology, in particular to a method for predicting ligand-protein interaction based on quantum computing.
Background
Quantum computing has been applied as an emerging technology for recent decades in the fields of finance, chemistry, biology, pharmacy, materials, etc. In classical machine learning, convolutional neural networks and hidden markov models are very important and are not powerful in many fields. With the advent of quantum computers, the potential of quantum computing has been gradually explored, and these classical algorithms are quantized, which has both practical significance and practical value. Quantum convolutional neural networks have been proposed in recent years, which have great potential in image processing and do not exhibit barren plateau phenomena; classical hidden markov has achieved a great deal of success in molecular structure prediction, and its corresponding quantum hidden markov has a great potential in structure prediction, similar to classical.
In the field of biopharmaceuticals, the discovery of a new drug consists of several parts, firstly, determining the target point of drug action, then designing or searching for effective compound molecules according to the spatial structure of the target point, and finally testing the physicochemical properties, metabolic properties and toxicological properties of the screened compound molecules one by one, selecting the drug meeting the requirements, and entering preclinical research and development. These processes are expensive and complex, involving large amounts of molecular data, which is a great burden on classical computers. In addition, the topology of the surface of the receptor protein is very complex and various, and the physical and chemical properties are also different, so that the positions capable of being combined with small molecular drugs and generating biological effects only occupy a very small part, and the finding of the proper binding site of the receptor protein becomes very difficult.
Therefore, it is necessary to develop an algorithm to predict these targets as well as potentially potent drug molecules with the acceleration advantage of quantum computing.
At present, many enterprises at home and abroad offer molecular docking services for predicting interaction between ligands and receptors, including binding modes and binding affinities thereof. Drug design involves molecular docking techniques, aided by computer, using linear regression, machine learning, neural networks, or other statistical techniques, by adapting experimental affinities to the calculated interactions between small molecules and targets. The method can predict the affinity before synthesizing the compound, so that only one compound is theoretically needed to be synthesized, thereby saving a great deal of time and cost and having great application prospect. In terms of ligand-receptor binding pattern prediction studies, the use of convolutional neural networks to predict drug-protein interactions has been studied with some feasibility, but the study was only performed with known binding sites. Because of the large volume of protein and the complex spatial structure, the traditional computer is difficult to research and predict based on the 3D spatial structure.
Disclosure of Invention
The invention aims to solve the problems of large calculation amount, large protein volume, complex space structure and difficult prediction which cannot be solved by the traditional computer, introduces quantum calculation into the research of the interaction prediction of ligand-protein, and realizes the synchronous prediction of the ligand-protein interaction and protein binding sites by means of the acceleration advantage of the quantum calculation.
The aim of the invention is realized by the following technical scheme:
a method of predicting ligand-protein interactions based on quantum computing is provided, the method comprising:
obtaining a 3D structure of the ligand drug molecule and the protein molecule;
predicting the 3D structure of the protein molecule by using a quantum hidden Markov model to obtain a possible binding site on the protein molecule;
docking the ligand drug molecule with the binding site and extracting features affecting ligand-protein docking using a quantum convolutional neural network;
scoring the docking process according to the characteristics affecting ligand-protein docking, and predicting the effect of ligand-protein interaction.
Specifically, the 3D structure of the ligand drug molecule and the protein molecule is obtained, comprising:
obtaining a three-dimensional crystal structure of the protein molecule by means of X-ray crystallography or NMR spectroscopy;
the 3D structure of a large number of ligand drug molecules is obtained by means of a large database of small molecule 3D structures.
Specifically, the method for obtaining the 3D structure of the ligand drug molecule and the protein molecule further comprises the following steps:
repairing errors generated in the protein analysis process, carrying out hydrogenation protonization on the three-dimensional crystal structure of protein molecules before butt joint, and marking local electrical property;
the 3D structure of the ligand drug molecule is subjected to energy minimization.
Specifically, the predicting the 3D structure of the protein molecule using a quantum hidden markov model includes:
learning the amino acid residue paths of known proteins in a three-dimensional space by using a de novo algorithm and a folding identification method to obtain three-dimensional path information of each folded amino acid;
encoding the atomic positions within each amino acid molecular lattice, placing each atom at a point in the amino acid molecular lattice;
each point on the amino acid molecular lattice can be moved to its one of its neighbors by using basis vectors, each as one state in a quantum hidden markov model;
learning the connection path of amino acids predicts the complex structure of the protein.
Specifically, prior to interfacing the ligand drug molecule with the binding site, further comprising:
the 3D structure of each drug molecule is encoded, each position encodes an atom in the corresponding molecule, the connecting edges correspond to chemical bonds between atoms in the molecule, and the drug molecule is described as a molecular diagram with atoms as nodes.
Specifically, interfacing the ligand drug molecule with the binding site comprises:
and docking the molecular diagram with a predicted binding site possibly existing on the protein molecule to obtain a ligand-protein connection structure diagram.
Specifically, the use of the quantum convolutional neural network to extract features that affect ligand-protein docking includes:
features affecting the ligand-protein interaction process are extracted using a quantum convolutional neural network.
Specifically, the scoring the docking process according to the characteristics affecting ligand-protein docking, including:
evaluating the effect of a certain docking using a scoring function, the effect comprising binding mode and affinity;
after multiple docking, ranking the scores for each binding site and drug molecule docked thereto;
a suitable molecule is selected in combination with human judgment as an alternative target for further docking testing.
In particular, the features include electrostatic interactions, hydrogen bonding interactions, hydrophobic interactions, and van der Waals interactions.
In particular, the case of docking is characterized using a molecular descriptor module.
It should be further noted that the technical features corresponding to the above options may be combined with each other or replaced to form a new technical scheme without collision.
Compared with the prior art, the invention has the beneficial effects that:
(1) According to the invention, based on the related information of molecular docking description of a quantum convolution neural network, a quantum hidden Markov model is used for predicting possible binding sites on receptor proteins, quantum computing is combined with the field of biological medicine, the difficulty of 3D space structure prediction of the whole protein of a classical computer is solved by utilizing the quantum hidden Markov and the quantum acceleration capability of the quantum hidden Markov, and simultaneously, the prediction of ligand-protein interaction is carried out under the condition of unknown binding sites, so that the synchronous prediction of molecular docking and protein structure is realized.
(2) Compared with a classical convolutional neural network, the molecular docking data are processed by using the quantum convolutional neural network, so that a barren plateau phenomenon does not occur, and the defect that the classical convolutional neural network cannot train aiming at a large problem is overcome.
(3) And 3D structure data of a large number of drug molecules are obtained by means of a large database of small molecule 3D structures, and energy minimization treatment is carried out, so that the stability of the molecular structure is maximized, and the method is used for training a model.
(4) By learning the connection path of amino acids, the complex structure of the protein is predicted, and thus binding sites for molecular docking that may exist on the protein surface are predicted.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are needed in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of a method for predicting ligand-protein interactions based on quantum computation, according to an embodiment of the present invention;
FIG. 2 is a roadmap of molecular docking and synchronous prediction of protein structure as shown in an embodiment of the invention;
FIG. 3 is a schematic flow diagram illustrating predicted binding sites and docking in accordance with an embodiment of the present invention;
FIG. 4 is a diagram of a quantum convolutional neural network architecture shown in an embodiment of the present invention;
FIG. 5 is a diagram showing the structure of the atomic position encoding within the lattice of an amino acid molecule according to an embodiment of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made apparent and fully understood from the accompanying drawings, in which some, but not all embodiments of the invention are shown. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
In addition, the technical features of the different embodiments of the present invention described below may be combined with each other as long as they do not collide with each other.
The invention mainly describes the related information of molecular docking based on a quantum convolution neural network, predicts the possible binding sites on the receptor protein by using a quantum hidden Markov model, realizes synchronous prediction of molecular docking and protein structure, combines quantum calculation with the biological medicine field, and solves the difficulty of protein 3D structure prediction.
Before describing embodiments of the present application, some terms referred to in the present application will be explained first.
1. Quantum computing: and a mode of rapidly completing a specific calculation task by utilizing superposition and entanglement properties of quantum states.
2. Quantum convolutional neural networks: based on quantum computing theory, the variance of the gradient of the proposed convolutional neural network is not disappeared faster than that of the polynomial, which means that the quantum convolutional neural network does not have a barren plateau phenomenon.
3. Basis vector: in linear algebra, the base is the basic tool for describing and characterizing vector space. The basis of the vector space is a special subset of it, and the elements of the basis are called basis vectors.
4. Scoring function: function for evaluating rationality of a theoretically obtained receptor-ligand binding pattern. The aim is to select the complex structure in the natural state. The method can be used for evaluating the receptor-ligand binding affinity of molecular docking and virtual screening. Three categories can be distinguished: a physics-based scoring function, a knowledge-based scoring function, and an experience-based scoring function.
5. De novo algorithm: the ab initio method refers to a quantum chemical calculation method of directly solving schrodinger's equation based on the basic principle of quantum mechanics in quantum chemical calculation. The de novo calculation is characterized by inexperienced parameters and does not overly simplify the system. The calculations were performed using essentially the same method for the various chemical systems. The de novo calculation method comprises a Hartri-Fox method based on a Hartri-Fox equation, a post-Hartri-Fox method developed by introducing electronic correlation correction on the basis of the Hartri-Fox equation, a multi-state multi-reference state method and the like. The de novo calculation method is highly accurate but takes a long time compared to the semi-empirical method.
6. Folding identification method: the task of fold identification is to find the type of fold to which an unknown protein belongs. The structure and function of unknown proteins are further deduced from the superfamily and fold class to which they belong.
7. X-ray crystallography method: x-ray crystallography is a discipline that uses X-rays to study the arrangement of atoms in a crystal. The structure and function of macromolecules such as proteins and DNA can be understood using X-ray crystallography. Accurate information on molecular structure is a prerequisite for rational drug design and structural function-based research.
8. NMR spectroscopy method: NMR was nuclear magnetic resonance. Is a physical process that the magnetic moment is not zero, the spin energy level is subjected to the zeeman splitting under the action of an external magnetic field, and the resonance absorbs the radio frequency radiation with a certain frequency. Nuclear magnetic resonance spectroscopy is a branch of spectroscopy whose resonance frequency is in the radio frequency band and the corresponding transition is a transition of nuclear spins at the nuclear zeeman level.
Example 1
A method for predicting ligand-protein interactions based on quantum computing is provided, as shown in fig. 1, the method comprising:
obtaining a 3D structure of the ligand drug molecule and the protein molecule;
predicting the 3D structure of the protein molecule by using a quantum hidden Markov model to obtain a possible binding site on the protein molecule;
docking the ligand drug molecule with the binding site and extracting features affecting ligand-protein docking using a quantum convolutional neural network;
scoring the docking process according to the characteristics affecting ligand-protein docking, and predicting the effect of ligand-protein interaction.
Specifically, the 3D structure of the drug molecule and the 3D structure of the protein molecule are obtained by methods such as crystallography or spectroscopy for predictive training. The basic method of protein structure prediction is then combined with a quantum hidden Markov model, and the complex structure of the protein is predicted by learning the connection path of amino acids, so that the possible binding sites for molecular docking on the surface of the protein are predicted.
Further, as shown in fig. 2, the ligand drug molecules are docked to the binding sites while predicting the binding sites, the situation of molecular docking is characterized using a quantum convolutional neural network, and features representing the situation of ligand-protein docking are extracted.
And finally, providing a scoring function according to the extracted characteristics to score the molecular docking, wherein the scoring is higher, the effect of the docking is better, and the drug molecules with good docking effect are selected.
According to the invention, based on the related information of molecular docking description of a quantum convolution neural network, a quantum hidden Markov model is used for predicting possible binding sites on receptor proteins, so that synchronous prediction of molecular docking and protein structure is realized, and meanwhile, quantum calculation is combined with the field of biological medicine, so that the difficulty in predicting the 3D structure of the protein is solved.
Example 2
Based on example 1, a method for predicting ligand-protein interactions based on quantum computation is provided, wherein the obtaining of the 3D structure of ligand drug molecules and protein molecules comprises:
obtaining a three-dimensional crystal structure of the protein molecule by means of X-ray crystallography or NMR spectroscopy;
the 3D structure of a large number of ligand drug molecules is obtained by means of a large database of small molecule 3D structures.
Further, the obtaining of the 3D structure of the ligand drug molecule and the protein molecule further comprises:
repairing errors generated in the protein analysis process, carrying out hydrogenation protonization on the three-dimensional crystal structure of the protein molecules before butt joint, marking local electrical property, and waiting for the next step;
the 3D structure of the ligand drug molecule is subjected to energy minimization.
Further, as shown in fig. 3, the predicting the 3D structure of the protein molecule using the quantum hidden markov model includes:
learning the amino acid residue paths of known proteins in a three-dimensional space by using a de novo algorithm and a folding identification method to obtain three-dimensional path information of each folded amino acid; a protein consists of multiple amino acids, each of which are tightly packed together to form a large lattice.
The atomic positions within the lattice of each amino acid molecule are encoded while taking into account the fixed distance between each atomic coordinate and the next atomic coordinate in three-dimensional space,
each point on the amino acid molecular lattice can be moved to its one of its neighbors by using basis vectors, each as one state in a quantum hidden markov model;
by learning the connection path of amino acids, the complex structure of the protein is predicted, and thus binding sites for molecular docking that may exist on the protein surface are predicted.
As shown in fig. 5, each atom is placed at a point in the amino acid molecular lattice, the primary structure of an amino acid is an amino acid sequence linked by peptide bonds, the primary structure of a protein is combined with the three-dimensional coordinates of amino acid residues in a hidden markov HMM to obtain a path of amino acid residues in the three-dimensional space of the protein from the first atom of the first amino acid molecular lattice to the last atom of the last amino acid molecular lattice, and each protein is depicted as a huge lattice of atoms linked.
Amino acid molecular lattice types are numerous and include simple cubic, body centered cubic, face centered cubic, and the like. FIG. 5 is a face-centered cubic lattice, as shown with the lattice center as the origin, encoding the position of each atom on the lattice, with the amino acid residue path in three dimensions shown as (V 1 ,V 2 ,…V n ) Wherein V is 1 And V n Three-dimensional space atomic coordinates of the first and last amino acids of the protein sequence, respectively. V (V) i-1 And V i Is the three-dimensional atomic coordinates of the adjacent amino acid molecular lattice, i=2, 3, …, n.
The atomic distance calculation is the same as the general lattice interatomic distance calculation method. By learning this amino acid residue pathway, the 3D structure of the protein can be reduced. Binding learning protein binding site information for known binding sites predicts binding sites that may exist on a new protein.
Further, prior to interfacing the ligand drug molecule with the binding site, further comprising:
the 3D structure of each drug molecule is encoded, each position encodes an atom in the corresponding molecule, the connecting edges correspond to chemical bonds between atoms in the molecule, and the drug molecule is described as a molecular diagram with atoms as nodes.
Further, interfacing the ligand drug molecule with the binding site, comprising:
and docking the molecular diagram with a predicted binding site possibly existing on the protein molecule to obtain a ligand-protein connection structure diagram.
Further, the use of the quantum convolutional neural network to extract features that affect ligand-protein docking, comprising:
features affecting the ligand-protein interaction process are extracted using a quantum convolutional neural network. Specifically, as shown in fig. 4, C in fig. 4 represents a convolution layer, P represents a pooling layer, and F represents a fully-connected layer. Input state ρ in As input, the state is sent through a circuit consisting of a series of convolutional layers and a pooling layer. The convolutional layer consists of two rows of parameterized double-qubit gates acting on alternating pairs of adjacent qubits. In each pooling layer half of the qubits are measured and the measurement control is applied to the gates of adjacent qubits. After multiple convolutional and pooled layers, the qcinn also contains a fully-connected layer that unifies the remaining qubits. Finally, the expected values of some Hermitian operators O are measured.
The first layer of convolution layer is used for fitting a ligand-protein molecule connection structure diagram, the second layer of pooling layer is used for extracting relevant characteristics affecting butt joint from the connection structure diagram, the third layer of convolution layer is used for further screening the characteristics, the fourth layer of pooling layer is used for reducing the dimension of the screened characteristics, the fifth layer of full-connection layer is used for scoring the butt joint, and rho out And outputting a scoring result.
Further, said scoring the docking process according to said characteristics affecting ligand-protein docking, comprising:
evaluating the effect of a certain docking using a scoring function, the effect comprising binding mode and affinity;
after multiple docking, ranking the scores for each binding site and drug molecule docked thereto;
a suitable molecule is selected in combination with human judgment as an alternative target for further docking testing.
Further, the features include electrostatic interactions, hydrogen bonding interactions, hydrophobic interactions, and van der Waals interactions.
Further, the case of docking is characterized using a molecular descriptor module.
Because of the large volume of protein, the complex space structure and high calculation complexity, the prediction of the 3D space structure of the whole protein is a huge difficult problem for a classical computer, and the difficulty of the 3D structure prediction of the protein is solved by utilizing the quantum hidden Markov and the quantum acceleration capability. Meanwhile, synchronous prediction of molecular docking and protein structure is realized, molecular docking data is processed by using a quantum convolutional neural network, and compared with a classical convolutional neural network, the method has the advantages that a barren plateau phenomenon can not occur, and the problem that the classical convolutional neural network cannot train aiming at a large-scale problem is solved.
Example 3
This embodiment has the same inventive concept as embodiment 1, and provides a storage medium having stored thereon computer instructions that, when executed, perform the steps of a quantum computation-based ligand-protein interaction prediction method of embodiment 1.
Based on such understanding, the technical solution of the present embodiment may be essentially or a part contributing to the prior art or a part of the technical solution may be embodied in the form of a software product stored in a storage medium, including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods of the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (Random Access Memory, RAM), a magnetic disk, or an optical disk, or other various media capable of storing program codes.
Example 4
The present embodiment also provides a terminal having the same inventive concept as embodiment 1, including a memory and a processor, the memory having stored thereon computer instructions executable on the processor, the processor executing the steps of a quantum computation-based ligand-protein interaction prediction method of embodiment 1 when the computer instructions are executed. The processor may be a single or multi-core central processing unit or a specific integrated circuit, or one or more integrated circuits configured to implement the invention.
The functional units in the embodiments provided in the present invention may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit.
The foregoing detailed description of the invention is provided for illustration, and it is not to be construed that the detailed description of the invention is limited to only those illustration, but that several simple deductions and substitutions can be made by those skilled in the art without departing from the spirit of the invention, and are to be considered as falling within the scope of the invention.
Claims (9)
1. A method of predicting ligand-protein interactions based on quantum computing, the method comprising:
obtaining a 3D structure of the ligand drug molecule and the protein molecule;
predicting the 3D structure of the protein molecule by using a quantum hidden Markov model to obtain a binding site existing on the protein molecule;
the predicting the 3D structure of the protein molecule using a quantum hidden markov model comprises:
learning the amino acid residue paths of known proteins in a three-dimensional space by using a de novo algorithm and a folding identification method to obtain three-dimensional path information of each folded amino acid;
encoding the atomic positions within each amino acid molecular lattice, placing each atom at a point in the amino acid molecular lattice;
each point on the amino acid molecular lattice is moved to its one of its neighbors by using basis vectors, each of which acts as one state in the quantum hidden markov model;
learning the connection path of amino acids, and predicting the complex structure of the protein;
obtaining a molecular diagram of ligand drug molecules, butting the molecular diagram with a predicted binding site existing on the obtained protein molecules to obtain a ligand-protein connection structure diagram, and extracting characteristics affecting ligand-protein butting by using a quantum convolutional neural network;
scoring the docking process according to the characteristics affecting ligand-protein docking, and predicting the effect of ligand-protein interaction.
2. A method of predicting ligand-protein interactions based on quantum computation according to claim 1, wherein said obtaining a 3D structure of a ligand drug molecule and a protein molecule comprises:
obtaining a three-dimensional crystal structure of the protein molecule by means of X-ray crystallography or NMR spectroscopy;
the 3D structure of a large number of ligand drug molecules is obtained by means of a large database of small molecule 3D structures.
3. A method of predicting ligand-protein interactions based on quantum computation according to claim 2, wherein the obtaining of the 3D structure of the ligand drug molecule and the protein molecule further comprises:
repairing errors generated in the protein analysis process, carrying out hydrogenation protonization on the three-dimensional crystal structure of protein molecules before butt joint, and marking local electrical property;
the 3D structure of the ligand drug molecule is subjected to energy minimization.
4. A method of predicting ligand-protein interactions based on quantum computing as recited in claim 1, further comprising, prior to interfacing the ligand drug molecule to the binding site:
the 3D structure of each drug molecule is encoded, each position encodes an atom in the corresponding molecule, the connecting edges correspond to chemical bonds between atoms in the molecule, and the drug molecule is described as a molecular diagram with atoms as nodes.
5. A method of quantum computing-based prediction of ligand-protein interactions according to claim 4, wherein interfacing the ligand drug molecule to the binding site comprises:
and butting the molecular diagram with a binding site existing on the predicted protein molecule to obtain a ligand-protein connection structure diagram.
6. The method of claim 5, wherein the extracting features affecting ligand-protein docking using a quantum convolutional neural network comprises:
features affecting the ligand-protein interaction process are extracted using a quantum convolutional neural network.
7. A method of predicting ligand-protein interactions based on quantum computing as recited in claim 6, wherein scoring the docking process based on the characteristics affecting ligand-protein docking comprises:
evaluating the effect of a certain docking using a scoring function, the effect comprising binding mode and affinity;
after multiple docking, ranking the scores for each binding site and drug molecule docked thereto;
in combination with human judgment, a suitable molecule is selected as an alternative target for further docking testing.
8. The method of claim 6, wherein the characteristics include electrostatic interactions, hydrogen bonding interactions, hydrophobic interactions, and van der waals interactions.
9. The method of claim 6, wherein the docking is characterized using a molecular descriptor module.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210079577.0A CN114446383B (en) | 2022-01-24 | 2022-01-24 | Quantum calculation-based ligand-protein interaction prediction method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210079577.0A CN114446383B (en) | 2022-01-24 | 2022-01-24 | Quantum calculation-based ligand-protein interaction prediction method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114446383A CN114446383A (en) | 2022-05-06 |
| CN114446383B true CN114446383B (en) | 2023-04-21 |
Family
ID=81369126
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210079577.0A Active CN114446383B (en) | 2022-01-24 | 2022-01-24 | Quantum calculation-based ligand-protein interaction prediction method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114446383B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115497554A (en) * | 2022-09-29 | 2022-12-20 | 电子科技大学 | Quantum system for predicting affinity between pharmaceutical compound and action target |
| CN115691654B (en) * | 2023-01-03 | 2023-05-19 | 北京中科弧光量子软件技术有限公司 | Method for predicting antibacterial peptide of quantum gate-controlled circulating neural network based on fewer parameters |
| CN116452101B (en) * | 2023-06-09 | 2023-08-25 | 淄博市中心医院 | Intelligent anesthesia department medicine distribution charging method and system |
| CN118072812B (en) * | 2024-04-18 | 2024-07-02 | 西安交通大学医学院第一附属医院 | Method for simulating protein-micromolecular dynamics based on quantum computation |
Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101970052A (en) * | 2007-12-21 | 2011-02-09 | 格莱坎生物科学股份有限公司 | Design and selection of medicaments that modulate the function and activity of interleukin 13 |
| CN103103238A (en) * | 2005-08-18 | 2013-05-15 | Ambrx公司 | Method of producing polypeptide with selected amino acids antibodies or antibody fragment at specific position in cells |
| CN105566469A (en) * | 2016-01-19 | 2016-05-11 | 上海交通大学 | Pear fruitlet gene and application thereof to adjustment of plant fruit size |
| CN107038348A (en) * | 2017-05-04 | 2017-08-11 | 四川大学 | Drug targets Forecasting Methodology based on protein ligands interaction finger-print |
| CN109021062A (en) * | 2018-08-06 | 2018-12-18 | 倍而达药业（苏州）有限公司 | A kind of screening technique of tumour neoantigen |
| CN109036580A (en) * | 2018-07-06 | 2018-12-18 | 华东师范大学 | The affine force prediction method of protein ligand based on interaction energy item and machine learning |
| KR20190030365A (en) * | 2017-09-14 | 2019-03-22 | 고려대학교 세종산학협력단 | Protein-ligand docking prediction method using quantum mechanics calculation and solvation effect |
| CN109671472A (en) * | 2017-11-23 | 2019-04-23 | 周林祥 | Electronic structure Computer-Aided Drug Design system and method |
| US11080607B1 (en) * | 2020-12-16 | 2021-08-03 | Ro5 Inc. | Data platform for automated pharmaceutical research using knowledge graph |
| CN113327644A (en) * | 2021-04-09 | 2021-08-31 | 中山大学 | Medicine-target interaction prediction method based on deep embedding learning of graph and sequence |
| CN113593633A (en) * | 2021-08-02 | 2021-11-02 | 中国石油大学(华东) | Drug-protein interaction prediction model based on convolutional neural network |
| WO2021218791A1 (en) * | 2020-04-29 | 2021-11-04 | 中国科学院上海药物研究所 | Prediction method and device for ligand-protein interaction |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CA2571956A1 (en) * | 2004-07-09 | 2006-02-16 | Wyeth | Methods and systems for predicting protein-ligand coupling specificities |
| KR101113903B1 (en) * | 2010-03-31 | 2012-03-06 | (주)퀀텀바이오솔루션즈 | Protein-ligand docking predinction method comprising quantum mechanical scoring |
| US20200342953A1 (en) * | 2019-04-29 | 2020-10-29 | International Business Machines Corporation | Target molecule-ligand binding mode prediction combining deep learning-based informatics with molecular docking |
-
2022
- 2022-01-24 CN CN202210079577.0A patent/CN114446383B/en active Active
Patent Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103103238A (en) * | 2005-08-18 | 2013-05-15 | Ambrx公司 | Method of producing polypeptide with selected amino acids antibodies or antibody fragment at specific position in cells |
| CN101970052A (en) * | 2007-12-21 | 2011-02-09 | 格莱坎生物科学股份有限公司 | Design and selection of medicaments that modulate the function and activity of interleukin 13 |
| CN105566469A (en) * | 2016-01-19 | 2016-05-11 | 上海交通大学 | Pear fruitlet gene and application thereof to adjustment of plant fruit size |
| CN107038348A (en) * | 2017-05-04 | 2017-08-11 | 四川大学 | Drug targets Forecasting Methodology based on protein ligands interaction finger-print |
| KR20190030365A (en) * | 2017-09-14 | 2019-03-22 | 고려대학교 세종산학협력단 | Protein-ligand docking prediction method using quantum mechanics calculation and solvation effect |
| CN109671472A (en) * | 2017-11-23 | 2019-04-23 | 周林祥 | Electronic structure Computer-Aided Drug Design system and method |
| CN109036580A (en) * | 2018-07-06 | 2018-12-18 | 华东师范大学 | The affine force prediction method of protein ligand based on interaction energy item and machine learning |
| CN109021062A (en) * | 2018-08-06 | 2018-12-18 | 倍而达药业（苏州）有限公司 | A kind of screening technique of tumour neoantigen |
| WO2021218791A1 (en) * | 2020-04-29 | 2021-11-04 | 中国科学院上海药物研究所 | Prediction method and device for ligand-protein interaction |
| US11080607B1 (en) * | 2020-12-16 | 2021-08-03 | Ro5 Inc. | Data platform for automated pharmaceutical research using knowledge graph |
| CN113327644A (en) * | 2021-04-09 | 2021-08-31 | 中山大学 | Medicine-target interaction prediction method based on deep embedding learning of graph and sequence |
| CN113593633A (en) * | 2021-08-02 | 2021-11-02 | 中国石油大学(华东) | Drug-protein interaction prediction model based on convolutional neural network |
Non-Patent Citations (5)
| Title |
|---|
| Xiaoyu Li等."Machine learning study of the relationship between the geometric and entropy discord".《Europhysics Letters》.2019,第127卷(第2期),第1-7页. * |
| Zhiqi Xiao等."Highly-Confident Protein Interactome Prediction Via Variational Autoencode".《2021 IEEE International Conference on Systems,man,and Cybernetics》.2022,第958-963页. * |
| 宋静林等."药物SAHA的潜在靶标预测及分子对接研究".《计算机与应用化学》.2013,第30卷(第1期),第97-101页. * |
| 李晓瑜等."基于量子条件主方程的隐马尔可夫模型".《电子科技大学学报》.2021,第50卷(第5期),第644-649页. * |
| 郭盛等."基于网络药理学与分子对接技术的补肺活血胶囊用于新型冠状病毒肺炎（COVID-19）恢复期治疗的分子机制研究".《中草药》.2020,第51卷(第9期),第2307-2316页. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114446383A (en) | 2022-05-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114446383B (en) | Quantum calculation-based ligand-protein interaction prediction method | |
| Rackers et al. | Tinker 8: software tools for molecular design | |
| Blaabjerg et al. | Rapid protein stability prediction using deep learning representations | |
| Labute | A widely applicable set of descriptors | |
| Jin et al. | Antibody-antigen docking and design via hierarchical structure refinement | |
| Bonella et al. | Theory and methods for rare events | |
| EP4158640A1 (en) | Systems and methods for determining molecular properties with atomic-orbital-based features | |
| US7890313B2 (en) | Method and apparatus for analysis of molecular combination based on computations of shape complementarity using basis expansions | |
| Sridharan et al. | Modern machine learning for tackling inverse problems in chemistry: molecular design to realization | |
| Pyrkov et al. | Quantum computing for near-term applications in generative chemistry and drug discovery | |
| JP5211458B2 (en) | Method and apparatus for virtual screening of compounds | |
| Pei et al. | Fabind: Fast and accurate protein-ligand binding | |
| Baldi et al. | A machine learning strategy for protein analysis | |
| Dhakal et al. | Predicting protein-ligand binding structure using E (n) Equivariant graph neural networks | |
| Hofstetter et al. | Graph-convolutional neural networks for (QM) ML/MM molecular dynamics simulations | |
| Sensoy et al. | Computational studies of G protein-coupled receptor complexes: Structure and dynamics | |
| Tan et al. | Cross-gate mlp with protein complex invariant embedding is a one-shot antibody designer | |
| CN116646029B (en) | Quantum approximation optimization algorithm-based molecular docking method, device, medium and equipment | |
| EP4352736A1 (en) | Adversarial framework for molecular conformation space modeling in internal coordinates | |
| CN115602252A (en) | Application of quantum computing algorithm and/or quantum computing device in development of improved new drugs | |
| Pandey et al. | Multibody molecular docking on a quantum annealer | |
| Boisson et al. | Parallel multi-objective algorithms for the molecular docking problem | |
| Thomas et al. | In silico protein engineering: Methods and Tools | |
| Du et al. | From Interatomic Distances to Protein Tertiary Structures with a Deep Convolutional Neural Network | |
| Wang | Force field development and nanoreactor chemistry |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |