Method and system for crystal resolution
By using a graph neural network-based method to automatically identify and resolve crystal structures, the problems of manual intervention and identification errors in the crystal resolution process are solved, and fast and accurate crystal resolution is achieved.
Patent Information
- Application Number
- CN202411053756.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-01
- Publication Date
- 2024-11-19
- Estimated Expiration
- 2044-08-01
AI Technical Summary
Crystal analysis requires extensive human intervention and specialized knowledge, is time-consuming and labor-intensive, and is prone to identification errors, making it difficult to achieve fast and accurate automated analysis.
By employing a graph neural network-based approach, the coordinates and element types of non-hydrogen atoms are determined by acquiring the electron density map of the crystal, a geometric relationship diagram is constructed, and the graph neural network module is used for automated identification and hydrogenation operations, thereby achieving automatic analysis of the crystal structure.
The automation of crystal analysis has been achieved, which significantly improves the analysis speed and accuracy, reduces human intervention and identification errors, and greatly enhances the efficiency of crystal analysis.
Smart Images

Figure CN118983014A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of artificial intelligence technology, and more specifically, to a method and system for crystal analysis. Background Art
[0002] The observation of the microscopic molecular structure of matter is a key factor in promoting scientific discovery. Crystal analysis is one of the important technologies, which can analyze the structure of crystals in detail, which is impossible to achieve with any other spectral analysis method. Crystal analysis plays a vital role in chemical research. In the history of more than 100 years, many Nobel Prizes in Chemistry have been closely related to crystal analysis. Nowadays, the research on new materials and new drugs has received widespread attention, involving the synthesis of a large number of new substances. Crystal analysis will be widely used in the structural analysis of newly synthesized substances, and has almost become a daily research in related fields. However, crystal analysis requires crystallographers to manually complete the analysis task through repeated trial and error with quite professional crystallographic knowledge. This is very time-consuming and labor-intensive for crystal analysis, a task that has a large demand in daily research. Therefore, a fast and accurate crystal analysis method and system is urgently needed. Summary of the invention
[0003] It is to be understood that both the foregoing general description and the following detailed description of the present invention are exemplary and explanatory and are intended to provide further explanation of the invention as claimed.
[0004] According to one aspect of the present invention, a method for crystal analysis is provided, comprising: obtaining an electron density map of a crystal, the electron density map being generated by phase analyzing diffraction data of the crystal; determining the coordinates of each non-hydrogen atom of the crystal based on the electron density peak of the electron density map; constructing a geometric relationship graph of the non-hydrogen atoms based on the coordinates of each non-hydrogen atom of the crystal; and determining the elemental type of each non-hydrogen atom of the crystal based on the geometric relationship graph of the non-hydrogen atoms using a non-hydrogen atom identification module based on a graph neural network.
[0005] According to another aspect of the present invention, a method for crystal analysis is provided, comprising: obtaining the coordinates and element type of each non-hydrogen atom of the crystal; constructing a geometric relationship diagram of equivalent atoms based on the coordinates and element type of each non-hydrogen atom of the crystal, wherein the equivalent atoms include non-hydrogen atoms and equivalent atoms; using a graph neural network-based hydrogenation module to determine the type of hydrogenation operation corresponding to each non-hydrogen atom of the crystal based on the geometric relationship diagram of equivalent atoms; and determining the coordinates of each hydrogen atom of the crystal based on the type of hydrogenation operation corresponding to each non-hydrogen atom of the crystal.
[0006] According to yet another aspect of the present invention, a system for crystal analysis is provided, comprising a device for executing any one of the methods described above.
[0007] The method and system for crystal analysis according to the embodiments of the present invention can achieve fast and accurate crystal analysis. BRIEF DESCRIPTION OF THE DRAWINGS
[0008] The accompanying drawings are included to provide a further understanding of the present invention, and are incorporated into and constitute a part of this application. The accompanying drawings illustrate embodiments of the present invention and together with the specification serve to explain the principles of the present invention. In the accompanying drawings:
[0009] Figure 1 is a flow chart of a method for crystal analysis according to one embodiment of the present invention;
[0010] Figure 2 is a flow chart of a method for constructing a geometric relationship diagram of non-hydrogen atoms according to one embodiment of the present invention;
[0011] Figure 3 is a flowchart of a method for training a non-hydrogen atom recognition module based on a graph neural network according to an embodiment of the present invention;
[0012] Figure 4 is a flow chart of a method for crystal analysis according to another embodiment of the present invention;
[0013] Figure 5 is a flow chart of a method for constructing a geometric relationship graph of equivalent atoms according to one embodiment of the present invention;
[0014] Figure 6 is a flow chart of a method for training a graph neural network-based hydrogenation module according to one embodiment of the present invention;
[0015] Figure 7 is a flow chart of a method for crystal analysis according to yet another embodiment of the present invention;
[0016] Figure 8 is a flow chart of a method for determining a hydroprocessing operation type according to one embodiment of the present invention;
[0017] Fig.9A is a schematic diagram of an electron density map and electron density peaks according to one embodiment of the present invention;
[0018] Fig. 9B is a schematic diagram of coordinates and element types of non-hydrogen atoms of a crystal according to one embodiment of the present invention;
[0019] Fig.10 is a block diagram of a system for crystal analysis according to one embodiment of the present invention;
[0020] Fig.11 is a block diagram of a system for crystal analysis according to another embodiment of the present invention;
[0021] Fig.12 is a block diagram of a system for crystal analysis according to yet another embodiment of the present invention;
[0022] Fig.13 is a result graph of the performance evaluation of crystal analysis; and
[0023] Fig.14 This is a comparison chart of crystal analysis applications. DETAILED DESCRIPTION
[0024] Embodiments of the present invention will now be described in detail with reference to the accompanying drawings, but the present invention is not limited thereto but only by the claims. In the accompanying drawings, for illustrative purposes, the sizes of some of the elements may be exaggerated and not drawn to scale. Wherever possible, the same reference numerals will be used in all drawings to represent the same or similar parts.
[0025] Although the terms used in the present invention are selected from well-known and commonly used terms, some of the terms mentioned in the present invention specification may be selected by the applicant at his or her discretion, and their detailed meanings are explained in the relevant parts of the description of this article. In addition, it is required to understand the present invention not only by the actual terms used, but also by the meaning contained in each term.
[0026] In the description provided herein, numerous specific details are set forth. However, it should be understood that embodiments of the present invention may be practiced without these specific details. In other examples, well-known methods, structures, and techniques are not shown in detail to avoid obscuring an understanding of the present invention.
[0027] At present, crystal analysis requires crystallographers to manually complete the analysis with professional crystallographic knowledge, and there is no fully automated tool. Crystal analysis generally uses X-rays to irradiate crystals to generate diffraction data. After phase analysis of the diffraction data, the diffraction data is restored to the electron density in the unit cell through Fourier transform. At this time, crystallographers need to manually identify non-hydrogen atoms in the electron density, that is, identify a certain position coordinate (x, y, z) in 3D space corresponding to a certain type of non-hydrogen atom (C, N, O, etc.) according to the morphology of the electron density. This identification process requires professional crystallographic knowledge and repeated attempts to verify the correctness, and after the identification is completed, the atomic coordinates need to be refined to better fit the theoretical diffraction data of the current structure and the real diffraction data obtained by experimental measurement. If it is observed that the fitting effect of the theoretical diffraction data and the real diffraction data is very poor after refinement, this is usually caused by the error in the identification of atoms in the previous step, so the wrongly identified atoms need to be re-identified. Therefore, crystallographers need to use professional crystallographic tools and rely on professional crystallographic knowledge to repeatedly identify and refine non-hydrogen atoms to obtain the correct structure, which is very time-consuming and human resources. Since crystal analysis is in great demand in daily research in related fields, identification by crystallographers is very time-consuming and labor-intensive, and there are many identification errors.
[0028] In view of the above problems, the present application proposes a method and system for crystal analysis.
[0029] Figure 1 FIG. 1 is a flow chart of a method 100 for crystal analysis according to an embodiment of the present invention. The method can be used for single crystals or for polycrystalline crystals.
[0030] At step 102, an electron density map of a crystal may be acquired. Fig.9A An example of an electron density map is shown on the left. Fig.9A The electron density map shown is in a two-dimensional form, and it should be understood that the electron density map actually reflects the three-dimensional spatial information of the crystal structure. The electron density map can be generated by phase parsing the diffraction data of the crystal. The diffraction data of the crystal may include information such as diffraction angle, diffraction intensity, structure factor, space group and symmetry, lattice parameters, etc. For polycrystals, diffraction data can be obtained from polycrystals by serial femtosecond crystallography, 3D electron diffraction, etc. For single crystals, in addition to the aforementioned methods, diffraction data can also be obtained from single crystals by methods such as X-rays. The diffraction data of the crystal can also be obtained from a database. In order to convert the diffraction data of the crystal into an electron density map, phase parsing can be achieved by crystallographic tools such as shelxs, shelxt, and charge flipping.
[0031] At step 104 , the coordinates of each non-hydrogen atom of the crystal may be determined based on the electron density peaks of the electron density map. Fig.9A The right side shows the electron density peaks determined according to the electron density map, each corresponding to a non-hydrogen atom. Since there are multiple electron density peaks in the electron density map, each electron density peak corresponds to a non-hydrogen atom, so the coordinates of each non-hydrogen atom can be determined by finding the coordinates of each electron density peak. Specifically, the coordinates of each electron density peak can be determined by finding the local maximum of the electron density map and then determining the coordinates corresponding to the position of each maximum. Although there are also electron density peaks corresponding to hydrogen atoms in the electron density map, since the peak value corresponding to hydrogen atoms is lower, the electron density peak corresponding to hydrogen atoms can be excluded according to the peak value compared to the electron density peak corresponding to non-hydrogen atoms.
[0032] At step 106, a geometric relationship graph of non-hydrogen atoms may be constructed based on the coordinates of each non-hydrogen atom of the crystal. The geometric relationship graph of non-hydrogen atoms may include the coordinates of each non-hydrogen atom and the connection relationship between the non-hydrogen atoms.
[0033] In one embodiment, step 106 may include steps 202 to 204, such as Figure 2 shown.
[0034] At step 202, the distance between each atom of all non-hydrogen atoms and each other atom may be calculated. For example, the distance between each non-hydrogen atom and each other non-hydrogen atom may be calculated based on the coordinates of each non-hydrogen atom.
[0035] At step 204, each pair of atoms in all non-hydrogen atoms whose distance is less than a distance threshold may be connected as nodes to construct a geometric relationship graph of non-hydrogen atoms. The distance threshold may be set as required.
[0036] return Figure 1 At step 108, a non-hydrogen atom identification module based on a graph neural network and a geometric relationship diagram of non-hydrogen atoms can be used to determine the element type of each non-hydrogen atom of the crystal. In some applications, for a crystal without hydrogen atoms, the coordinates and element type of each non-hydrogen atom determined according to the above method constitute a complete crystal structure. In other applications, the coordinates and element type of each non-hydrogen atom determined according to the above method constitute the skeleton of the crystal, which can be used for preliminary crystal structure analysis.
[0037] In one embodiment, bonding between non-hydrogen atoms may be calculated based on covalent radii corresponding to the element types of the non-hydrogen atoms.
[0038] Instead of requiring crystallographic experts to use crystallographic knowledge to perform crystal analysis through repeated manual attempts, by using the above-mentioned method for crystal analysis, the present application can achieve automatic crystal analysis without manual identification by human experts. In addition, the present application can achieve a speed thousands of times faster than that of human experts and a higher accuracy rate, which can greatly free up time and labor from tedious manual crystal analysis.
[0039] Figure 3 3 is a flowchart of a method 300 for training a non-hydrogen atom recognition module based on a graph neural network according to an embodiment of the present invention. The non-hydrogen atom recognition module based on a graph neural network can be trained by the method 300.
[0040] At step 302, a data set may be received. The data set may include a geometric relationship diagram of non-hydrogen atoms of each crystal in a plurality of crystals, and a true value of the element type of the non-hydrogen atoms. The plurality of crystals may include single crystals, or may include polycrystals. The geometric relationship diagram of the non-hydrogen atoms of each crystal may be constructed based on the diffraction data of the crystal by a method similar to steps 102 to 106.
[0041] At step 304, a non-hydrogen atom identification module based on a graph neural network can be used to generate an element type prediction result for each non-hydrogen atom of each crystal in a plurality of crystals based on the data set. The non-hydrogen atom identification module based on a graph neural network can be initialized based on a graph neural network model, and the nodes (non-hydrogen atoms) in the geometric relationship graph of the non-hydrogen atoms of each crystal are classified to predict the element type corresponding to each non-hydrogen atom. Since the graph neural network is particularly suitable for processing graph structure data, the non-hydrogen atom identification module based on the graph neural network can realize fast and accurate classification of nodes in the geometric relationship graph of non-hydrogen atoms.
[0042] At step 306, a loss function between a predicted result of the element type of each non-hydrogen atom of each crystal in the plurality of crystals and a true value may be calculated. The loss function may be a cross entropy loss, a mean square error, or the like.
[0043] At step 308, the parameters of the graph neural network-based non-hydrogen atom identification module can be updated based on the loss function and using a parameter updating method to obtain a trained graph neural network-based non-hydrogen atom identification module. The parameter updating method can be gradient backpropagation, gradient descent, momentum optimization, etc. By updating the parameters through the loss function and the parameter updating method, the performance of the module can be optimized.
[0044] By executing method 300, a trained graph neural network-based non-hydrogen atom recognition module can be obtained. The present application utilizes a graph neural network-based non-hydrogen atom recognition module to achieve rapid and accurate recognition of non-hydrogen atoms.
[0045] In crystal analysis, in addition to the identification of non-hydrogen atoms, another urgent problem to be solved is how to identify hydrogen atoms attached to non-hydrogen atoms. Due to the weak scattering ability of hydrogen atoms to X-rays, hydrogen atoms cannot be directly observed directly from the electron density map of crystal diffraction data in most cases. At present, crystallographers are also required to judge which non-hydrogen atoms to add hydrogen atoms based on professional crystallographic knowledge. Since hydrogen atoms are difficult to observe from electron density maps, it is very easy to make omission errors or addition errors, and crystallographers need to try repeatedly to correctly add hydrogen. In view of the above problems, the present application proposes another method and system for crystal analysis.
[0046] Figure 4 FIG. 4 is a flow chart of a method 400 for crystal analysis according to another embodiment of the present invention. The method can be used for single crystals or multi-crystals.
[0047] At step 402, the coordinates and element type of each non-hydrogen atom of the crystal may be obtained. Fig. 9B A schematic diagram of the coordinates and element type of each non-hydrogen atom representing the crystal is shown, wherein the connection relationship between the non-hydrogen atoms indicates the bonding that can be calculated according to the covalent radius corresponding to the element type of the non-hydrogen atom, but the information obtained in step 402 does not need to include the bonding information. In one embodiment, the coordinates and element type of each non-hydrogen atom of the crystal can be obtained from the diffraction data of the crystal. In one embodiment, the coordinates and element type of each non-hydrogen atom of the crystal can be determined based on the diffraction data of the crystal using a non-hydrogen atom identification module based on a graph neural network. In another embodiment, the coordinates and element type of each non-hydrogen atom of the crystal can be determined based on the diffraction data of the crystal by automated software. In yet another embodiment, the coordinates and element type of each non-hydrogen atom of the crystal can be determined based on the diffraction data of the crystal by manual identification. The diffraction data of the crystal can include information such as diffraction angle, diffraction intensity, structure factor, space group and symmetry, lattice parameters, etc. For polycrystals, diffraction data can be obtained from polycrystals by serial femtosecond crystallography, 3D electron diffraction, etc. For single crystals, in addition to the aforementioned methods, diffraction data can also be obtained from single crystals by X-rays and other methods. The diffraction data of crystals can also be obtained from the database.
[0048] At step 404 , a geometric relationship map of equivalent atoms may be constructed based on the coordinates and element type of each non-hydrogen atom of the crystal, wherein the equivalent atoms include non-hydrogen atoms and equivalent atoms.
[0049] In one embodiment, step 404 may include steps 502 to 506. Figure 5 shown.
[0050] At step 502, the symmetry and periodicity of the crystal can be used to determine the coordinates and element types of equivalent atoms within a predetermined distance of each non-hydrogen atom of the crystal. In chemical research, based on the symmetry and periodicity of the crystal, the position and element type of the equivalent atoms can be determined when the position and element type of the non-hydrogen atoms are known. Since the internal structure of the crystal is infinitely repeated, it is only necessary to determine the equivalent atoms within a predetermined distance of the non-hydrogen atoms. The predetermined distance can be set as needed.
[0051] At step 504, the distance between each equivalent atom in the equivalent atoms and each other equivalent atom can be calculated, wherein the equivalent atoms include non-hydrogen atoms and the determined equivalent atoms. Specifically, the distance between each equivalent atom and each other equivalent atom can be calculated based on the coordinates of each equivalent atom.
[0052] At step 506, each pair of equivalent atoms whose distance is less than a distance threshold can be used as nodes to connect edges to construct a geometric relationship graph of equivalent atoms. The distance threshold can be set as needed. The distance threshold can be the same as or different from the distance threshold used in constructing the geometric relationship graph of non-hydrogen atoms in step 204.
[0053] return Figure 4 At step 406, a hydrogenation module based on a graph neural network may be used to determine the type of hydrogenation operation corresponding to each non-hydrogen atom of the crystal based on the geometric relationship graph of equivalent atoms. The hydrogenation operation type may include, for example, no hydrogenation, adding 1 hydrogen atom to a carbon atom, adding 2 hydrogen atoms to a carbon atom, adding 3 hydrogen atoms to a carbon atom, adding 1 hydrogen atom to an oxygen atom, and the like.
[0054] At step 408, the coordinates of each hydrogen atom of the crystal may be determined based on the type of hydrogenation operation corresponding to each non-hydrogen atom of the crystal. When the type of hydrogenation operation is determined, the coordinates of the hydrogen atoms corresponding to the non-hydrogen atoms may be uniquely determined based on the bonding calculated based on the covalent radius corresponding to the element type of the non-hydrogen atom.
[0055] After determining the coordinates of the hydrogen atoms corresponding to each non-hydrogen atom of the crystal, in combination with the coordinates and element type of each non-hydrogen atom of the crystal obtained, the coordinates and element type of all atoms of the molecule in a real environment can be obtained. In one embodiment, the visualization of the crystal structure can be realized based on the coordinates and element type of all atoms. In another embodiment, a crystallographic information file (Crystallographic Information File, CIF) can be generated based on the coordinates and element type of all atoms. The CIF file contains crystal structure information and can be read and / or processed by computer software.
[0056] Instead of requiring crystallographic experts to use crystallographic knowledge to perform crystal analysis through repeated manual attempts, by using the above-mentioned method for crystal analysis, the present application can achieve automatic crystal analysis without manual identification by human experts. In addition, the present application can achieve a speed thousands of times faster than that of human experts and a higher accuracy rate, which can greatly free up time and labor from tedious manual crystal analysis.
[0057] Figure 6 6 is a flow chart of a method 600 for training a hydrogenation module based on a graph neural network according to an embodiment of the present invention. The hydrogenation module based on a graph neural network can be trained by the method 600.
[0058] At step 602, a data set may be received. The data set may include a geometric relationship map of equivalent atoms for each crystal in a plurality of crystals, and a true value of the hydrogenation type of the non-hydrogen atoms. The plurality of crystals may include single crystals, or may include polycrystals. The geometric relationship map of equivalent atoms for each crystal may be constructed based on the coordinates and element type of the non-hydrogen atoms of the crystal by a method similar to step 404.
[0059] At step 604, a hydrogenation module based on a graph neural network can be used to generate a prediction result of the hydrogenation operation type of each non-hydrogen atom of each crystal in a plurality of crystals based on the data set. The hydrogenation module based on a graph neural network can be initialized based on a graph neural network model, and the non-hydrogen atoms in the geometric relationship graph of equivalent atoms of each crystal are classified to predict the hydrogenation operation type corresponding to each non-hydrogen atom. Since graph neural networks are particularly suitable for processing graph structure data, the hydrogenation module based on a graph neural network can achieve fast and accurate classification of non-hydrogen atoms in the geometric relationship graph of equivalent atoms.
[0060] At step 606, a loss function between a predicted result and a true value of the hydrogenation operation type of each non-hydrogen atom of each crystal in the plurality of crystals may be calculated. The loss function may be a cross entropy loss, a mean square error, or the like.
[0061] At step 608, the parameters of the graph neural network-based hydrogenation module can be updated based on the loss function and using a parameter update method to obtain a trained graph neural network-based hydrogenation module. The parameter update method can be gradient backpropagation, gradient descent, momentum optimization, etc. By updating the parameters through the loss function and the parameter update method, the performance of the module can be optimized.
[0062] By executing method 600, a trained graph neural network-based hydrogenation module can be obtained. The present application utilizes a graph neural network-based hydrogenation module to achieve rapid and accurate determination of the type of hydrogenation operation of non-hydrogen atoms.
[0063] Figure 7 FIG. 7 is a flow chart of a method 700 for crystal analysis according to another embodiment of the present invention. The method can be used for single crystals or multi-crystals.
[0064] Steps 702 to 708 and reference Figure 1 Steps 102 to 108 in the described method 100 are the same. To avoid redundancy, the specific details of steps 702 to 708 are not described here.
[0065] Based on the determined coordinates and element types of the non-hydrogen atoms, in order to determine the complete crystal structure, the type of hydrogenation operation corresponding to each non-hydrogen atom can be further determined to determine the complete coordinates of all atoms including non-hydrogen atoms and hydrogen atoms.
[0066] At step 710, the type of hydrogenation operation corresponding to each non-hydrogen atom of the crystal may be determined. In one embodiment, the type of hydrogenation operation corresponding to each non-hydrogen atom of the crystal may be determined by manual identification based on the coordinates and element type of each non-hydrogen atom of the crystal determined in steps 704 and 708. In another embodiment, the type of hydrogenation operation corresponding to each non-hydrogen atom of the crystal may be determined using a graph neural network-based hydrogenation operation module based on the coordinates and element type of each non-hydrogen atom of the crystal determined in steps 704 and 708.
[0067] In one embodiment, step 710 may include steps 802 to 804. Figure 8 shown.
[0068] At step 802, a geometric relationship diagram of equivalent atoms can be constructed based on the coordinates and element type of each non-hydrogen atom of the crystal, wherein the equivalent atoms include non-hydrogen atoms and equivalent atoms. The geometric relationship diagram of equivalent atoms can include the coordinates of each equivalent atom and the connection relationship between the equivalent atoms.
[0069] In one embodiment, constructing a geometric relationship graph of equivalent atoms based on the coordinates and element type of each non-hydrogen atom of the crystal may include: utilizing the symmetry and periodicity of the crystal to determine the coordinates and element type of equivalent atoms of each non-hydrogen atom of the crystal within a predetermined distance, calculating the distance between each equivalent atom in the equivalent atoms to every other equivalent atom, wherein the equivalent atoms include the non-hydrogen atoms and the determined equivalent atoms, and connecting each pair of equivalent atoms in the equivalent atoms whose distance is less than a distance threshold as nodes to construct a geometric relationship graph of equivalent atoms.
[0070] At step 804, a graph neural network-based hydrogenation module may be used to determine the type of hydrogenation operation corresponding to each non-hydrogen atom of the crystal based on the geometric relationship graph of equivalent atoms. The hydrogenation operation type may include, for example, no hydrogenation, adding 1 hydrogen atom to a carbon atom, adding 2 hydrogen atoms to a carbon atom, adding 3 hydrogen atoms to a carbon atom, adding 1 hydrogen atom to an oxygen atom, and the like.
[0071] In one embodiment, the graph neural network-based hydrogenation module can be trained by the following steps: receiving a data set, the data set including a geometric relationship diagram of equivalent atoms of each crystal in a plurality of crystals, and the true value of the hydrogenation operation type of the non-hydrogen atom; using the graph neural network-based hydrogenation module to generate a prediction result of the hydrogenation operation type of each non-hydrogen atom in each crystal in a plurality of crystals based on the data set; calculating a loss function between the prediction result and the true value of the hydrogenation operation type of each non-hydrogen atom in each crystal in a plurality of crystals; and based on the loss function, using a parameter updating method to update the parameters of the graph neural network-based hydrogenation module to obtain a trained graph neural network-based hydrogenation module.
[0072] return Figure 7 At step 712, the coordinates of each hydrogen atom of the crystal may be determined based on the type of hydrogenation operation corresponding to each non-hydrogen atom of the crystal. When the type of hydrogenation operation is determined, the coordinates of the hydrogen atoms corresponding to the non-hydrogen atoms may be uniquely determined based on the bonding calculated based on the covalent radius corresponding to the element type of the non-hydrogen atom.
[0073] After determining the coordinates and element type of each non-hydrogen atom of the crystal and the coordinates of the hydrogen atom corresponding to each non-hydrogen atom, the coordinates and element types of all atoms of the molecule in the real environment can be obtained. In one embodiment, the visualization of the crystal structure can be achieved based on the coordinates and element types of all atoms. In another embodiment, a CIF file can be generated based on the coordinates and element types of all atoms.
[0074] Instead of requiring crystallographic experts to use crystallographic knowledge to perform crystal analysis through repeated manual attempts, by using the above-mentioned method for crystal analysis, the present application can achieve automatic crystal analysis, and can achieve faster speed and higher accuracy.
[0075] Fig.101 is a block diagram of a system 1000 for crystal analysis according to an embodiment of the present invention. System 1000 may include an electron density map acquisition module 1002, a non-hydrogen atom coordinate determination module 1004, a non-hydrogen atom geometric relationship diagram construction module 1006, and a non-hydrogen atom identification module 1008 based on a graph neural network. The electron density map acquisition module 1002 can be used to acquire an electron density map of a crystal, and the electron density map is generated by phase parsing the diffraction data of the crystal. The non-hydrogen atom coordinate determination module 1004 can be used to determine the coordinates of each non-hydrogen atom of the crystal based on the electron density peak of the electron density map. The non-hydrogen atom geometric relationship diagram construction module 1006 can be used to construct a non-hydrogen atom geometric relationship diagram based on the coordinates of each non-hydrogen atom of the crystal. The non-hydrogen atom identification module 1008 based on the graph neural network can be used to determine the element type of each non-hydrogen atom of the crystal based on the non-hydrogen atom geometric relationship diagram.
[0076] In one embodiment, the geometric relationship graph construction module 1006 of non-hydrogen atoms can be used to construct a geometric relationship graph of non-hydrogen atoms based on the coordinates of each non-hydrogen atom in the crystal through the following steps: calculating the distance between each atom among all non-hydrogen atoms and each other atom, and connecting each pair of atoms among all non-hydrogen atoms whose distance is less than a distance threshold as nodes to construct a geometric relationship graph of non-hydrogen atoms.
[0077] In one embodiment, the non-hydrogen atom identification module 1008 based on a graph neural network can be trained by the following steps: receiving a data set, the data set including a geometric relationship diagram of non-hydrogen atoms of each crystal in a plurality of crystals, and the true value of the element type of the non-hydrogen atoms; using the non-hydrogen atom identification module based on a graph neural network to generate a prediction result of the element type of each non-hydrogen atom in each crystal in a plurality of crystals based on the data set; calculating a loss function between the prediction result and the true value of the element type of each non-hydrogen atom in each crystal in a plurality of crystals; and updating the parameters of the non-hydrogen atom identification module based on the graph neural network using a parameter updating method based on the loss function to obtain a trained non-hydrogen atom identification module based on the graph neural network.
[0078] Fig.111 is a block diagram of a system 1100 for crystal analysis according to another embodiment of the present invention. The system 1100 may include a non-hydrogen atom acquisition module 1102, a geometric relationship graph construction module 1104 of equivalent atoms, a hydrogenation module 1106 based on a graph neural network, and a hydrogen atom coordinate determination module 1108. The non-hydrogen atom acquisition module 1102 can be used to obtain the coordinates and element type of each non-hydrogen atom of the crystal. The geometric relationship graph construction module 1104 of equivalent atoms can be used to construct a geometric relationship graph of equivalent atoms based on the coordinates and element type of each non-hydrogen atom of the crystal, wherein the equivalent atoms include non-hydrogen atoms and equivalent atoms. The hydrogenation module 1106 based on the graph neural network can be used to determine the type of hydrogenation operation corresponding to each non-hydrogen atom of the crystal based on the geometric relationship graph of equivalent atoms. The hydrogen atom coordinate determination module 1108 can be used to determine the coordinates of each hydrogen atom of the crystal based on the type of hydrogenation operation corresponding to each non-hydrogen atom of the crystal.
[0079] In one embodiment, the geometric relationship graph construction module 1104 of equivalent atoms can be used to construct a geometric relationship graph of equivalent atoms based on the coordinates and element type of each non-hydrogen atom of the crystal through the following steps: using the symmetry and periodicity of the crystal to determine the coordinates and element type of the equivalent atoms of each non-hydrogen atom of the crystal within a predetermined distance; calculating the distance between each equivalent atom in the equivalent atoms to each other equivalent atom, wherein the equivalent atoms include the non-hydrogen atoms and the determined equivalent atoms; and connecting each pair of equivalent atoms in the equivalent atoms whose distance is less than a distance threshold as nodes to construct a geometric relationship graph of equivalent atoms.
[0080] In one embodiment, the graph neural network-based hydrogenation module 1106 can be trained by the following steps: receiving a data set, the data set including a geometric relationship diagram of equivalent atoms of each crystal in a plurality of crystals, and the true value of the hydrogenation operation type of the non-hydrogen atom; using the graph neural network-based hydrogenation module to generate a prediction result of the hydrogenation operation type of each non-hydrogen atom in each of the plurality of crystals based on the data set; calculating a loss function between the prediction result and the true value of the hydrogenation operation type of each non-hydrogen atom in each of the plurality of crystals; and based on the loss function, using a parameter updating method to update the parameters of the graph neural network-based hydrogenation module to obtain a trained graph neural network-based hydrogenation module.
[0081] Fig.121 is a block diagram of a system 1200 for crystal analysis according to another embodiment of the present invention. The system 1200 may include an electron density map acquisition module 1202, a non-hydrogen atom coordinate determination module 1204, a non-hydrogen atom geometric relationship diagram construction module 1206, a non-hydrogen atom identification module based on a graph neural network 1208, a hydrogenation operation type determination module 1210, and a hydrogen atom coordinate determination module 1212. The electron density map acquisition module 1202 to the non-hydrogen atom identification module 1208 based on the graph neural network are connected to the reference Fig.10 The electron density map acquisition module 1002 to the non-hydrogen atom identification module 1008 based on the graph neural network in the described system 1000 are the same. To avoid redundancy, the specific details of the electron density map acquisition module 1202 to the non-hydrogen atom identification module 1208 based on the graph neural network are not described here. The hydrogenation operation type determination module 1210 can be used to determine the type of hydrogenation operation corresponding to each non-hydrogen atom of the crystal. The hydrogen atom coordinate determination module 1212 can be used to determine the coordinates of each hydrogen atom of the crystal based on the type of hydrogenation operation corresponding to each non-hydrogen atom of the crystal.
[0082] In one embodiment, the hydrogenation operation type determination module 1210 may include a geometric relationship graph construction module through equivalent atoms and a hydrogenation module based on a graph neural network. The geometric relationship graph construction module through equivalent atoms may be used to construct a geometric relationship graph through equivalent atoms based on the coordinates and element type of each non-hydrogen atom of the crystal, wherein the equivalent atoms include non-hydrogen atoms and equivalent atoms. The hydrogenation module based on the graph neural network may be used to determine the type of hydrogenation operation corresponding to each non-hydrogen atom of the crystal based on the geometric relationship graph through equivalent atoms.
[0083] In one embodiment, a geometric relationship graph construction module of equivalent atoms can be used to construct a geometric relationship graph of equivalent atoms based on the coordinates and element type of each non-hydrogen atom of the crystal through the following steps: utilizing the symmetry and periodicity of the crystal to determine the coordinates and element type of equivalent atoms of each non-hydrogen atom of the crystal within a predetermined distance; calculating the distance between each equivalent atom in the equivalent atoms to each other equivalent atom, wherein the equivalent atoms include the non-hydrogen atoms and the determined equivalent atoms; and connecting each pair of equivalent atoms in the equivalent atoms whose distance is less than a distance threshold as nodes to construct a geometric relationship graph of equivalent atoms.
[0084] In one embodiment, a graph neural network-based hydrogenation module is trained by the following steps: receiving a data set, the data set including a geometric relationship graph of equivalent atoms of each crystal in a plurality of crystals, and true values of the hydrogenation operation types of non-hydrogen atoms; using the graph neural network-based hydrogenation module to generate a prediction result of the hydrogenation operation type of each non-hydrogen atom in each crystal in a plurality of crystals based on the data set; calculating a loss function between the prediction result and the true value of the hydrogenation operation type of each non-hydrogen atom in each crystal in a plurality of crystals; and updating the parameters of the graph neural network-based hydrogenation module using a parameter updating method based on the loss function to obtain a trained graph neural network-based hydrogenation module.
[0085] Fig.13 It is a result diagram of the performance evaluation of crystal analysis. This application performs a performance evaluation on 10,000 single crystal diffraction data that are actually measured. Part (a) shows the performance of the non-hydrogen atom identification module in non-hydrogen atom identification. On many element types that almost cover the entire periodic table, the non-hydrogen atom identification module generally achieves quite high performance. Part (b) shows the performance of the hydrogenation module in determining the type of hydrogenation operation, and it has achieved high performance on a variety of common hydrogenation operation types. Part (c) shows the crystal prediction accuracy, that is, the proportion of crystals with all non-hydrogen atoms predicted correctly to all crystals. The non-hydrogen atom identification module of this application has achieved a very significant performance improvement in the problem of non-hydrogen atom identification compared with the only traditional crystallographic tool that currently has a certain non-hydrogen atom identification capability. Top-1 refers to the accuracy of the element type directly taking the highest probability of model prediction in this application, that is, the accuracy of the model directly predicting correctly. Top-2 refers to the accuracy of the element type directly taking the highest probability of model prediction in this application, which contains at most one erroneous atom, and the correct element type of the erroneous atom can be corrected by taking the element type with the second highest probability of model prediction. Parts (d) and (e) show that in terms of indicators reflecting the quality of analysis (residual (R1), goodness of fit (S)), the analysis results of the non-hydrogen atom identification module combined with the hydrogenation module are highly consistent with the analysis results of human experts. The analysis of this application is fully automatic, and the analysis speed is thousands of times that of human experts.
[0086] When the method of the present application was applied to crystal diffraction data provided by multiple published documents, the method of the present application found multiple errors in the crystal structure. Fig.14It is a comparison chart of crystal analysis application, with the crystal structure diagram shown in the published literature on the left and the non-hydrogen atoms and hydrogen atoms identified according to the method of this application on the right. As indicated by the dotted box, there are three typical errors in the crystal structure in the published literature, namely, non-hydrogen atom identification errors (part a), hydrogen atom addition errors (part b), and hydrogen atom addition omissions (part c). In addition, the residual (R1) of the comparison between the crystal structure identified by this application and the real experimental data is also reduced compared with the published literature.
[0087] Compared with crystallographic experts who use crystallographic knowledge to perform crystal analysis through repeated manual attempts, the automatic crystal analysis method implemented in this application through the use of artificial intelligence technology can complete the crystal analysis task without any professional knowledge, greatly improve the analysis speed, and ensure the accuracy of the analysis.
[0088] Further examples:
[0089] Example 1. A method for crystal analysis, comprising:
[0090] Obtaining an electron density map of a crystal, the electron density map being generated by phase resolving diffraction data of the crystal;
[0091] determining the coordinates of each non-hydrogen atom of the crystal based on the electron density peaks of the electron density map;
[0092] constructing a geometric relationship map of non-hydrogen atoms based on the coordinates of each non-hydrogen atom of the crystal; and
[0093] A non-hydrogen atom identification module based on a graph neural network is used to determine the element type of each non-hydrogen atom of the crystal based on a geometric relationship graph of the non-hydrogen atoms.
[0094] Example 2. The method of Example 1, wherein constructing a geometric relationship diagram of non-hydrogen atoms based on the coordinates of each non-hydrogen atom of the crystal comprises:
[0095] Calculate the distance from each atom to every other atom among all non-hydrogen atoms; and
[0096] Each pair of atoms in all non-hydrogen atoms whose distance is less than a distance threshold is taken as a node and connected to an edge to construct a geometric relationship graph of the non-hydrogen atoms.
[0097] Example 3. The method of Example 1, wherein the graph neural network-based non-hydrogen atom recognition module is trained by the following steps:
[0098] receiving a data set comprising a geometric relationship map of non-hydrogen atoms and true values of elemental types of the non-hydrogen atoms for each of a plurality of crystals;
[0099] Generate an element type prediction result of each non-hydrogen atom of each crystal in the plurality of crystals based on the data set using the graph neural network-based non-hydrogen atom identification module;
[0100] calculating a loss function between a predicted result and a true value of the element type of each non-hydrogen atom of each crystal in the plurality of crystals; and
[0101] Based on the loss function, a parameter updating method is used to update the parameters of the graph neural network-based non-hydrogen atom recognition module to obtain a trained graph neural network-based non-hydrogen atom recognition module.
[0102] Example 4. The method of Example 1 further comprises:
[0103] determining a type of hydrogenation operation corresponding to each non-hydrogen atom of the crystal; and
[0104] Based on the type of hydrogenation operation corresponding to each non-hydrogen atom of the crystal, the coordinates of each hydrogen atom of the crystal are determined.
[0105] Example 5. The method of Example 4, wherein determining the type of hydrogenation operation corresponding to each non-hydrogen atom of the crystal comprises:
[0106] constructing a geometric relationship diagram of equivalent atoms based on the coordinates and element type of each non-hydrogen atom of the crystal, wherein the equivalent atoms include non-hydrogen atoms and equivalent atoms; and
[0107] A graph neural network-based hydrogenation module is used to determine a type of hydrogenation operation corresponding to each non-hydrogen atom of the crystal based on a geometric relationship graph of the equivalent atoms.
[0108] Example 6. The method of Example 5, wherein constructing a geometric relationship diagram of equivalent atoms based on the coordinates and element type of each non-hydrogen atom of the crystal comprises:
[0109] Determine the coordinates and element type of the equivalent atom located within a predetermined distance of each non-hydrogen atom of the crystal by using the symmetry and periodicity of the crystal;
[0110] Calculating the distance between each equivalent atom and every other equivalent atom in the equivalent atoms, wherein the equivalent atoms include non-hydrogen atoms and the determined equivalent atoms; and
[0111] Each pair of equivalent atoms whose distance among the equivalent atoms is less than a distance threshold is taken as a node and connected to an edge, so as to construct a geometric relationship graph of the equivalent atoms.
[0112] Example 7. The method of Example 5, wherein the graph neural network-based hydrogenation module is trained by the following steps:
[0113] receiving a data set comprising a geometric relationship map of equivalent atoms and true values of hydrogenation types of non-hydrogen atoms for each crystal in a plurality of crystals;
[0114] Using the graph neural network-based hydrogenation module, based on the data set, to generate a prediction result of the hydrogenation operation type for each non-hydrogen atom of each crystal in the plurality of crystals;
[0115] calculating a loss function between a predicted result and a true value of the hydrogenation operation type for each non-hydrogen atom of each crystal in the plurality of crystals; and
[0116] Based on the loss function, a parameter updating method is used to update the parameters of the graph neural network-based hydrogenation module to obtain a trained graph neural network-based hydrogenation module.
[0117] Example 8. A method for crystal analysis, comprising:
[0118] Get the coordinates and element type of each non-hydrogen atom in the crystal;
[0119] constructing a geometric relationship diagram of equivalent atoms based on the coordinates and element type of each non-hydrogen atom of the crystal, wherein the equivalent atoms include non-hydrogen atoms and equivalent atoms;
[0120] Determining a type of hydrogenation operation corresponding to each non-hydrogen atom of the crystal based on a geometric relationship graph of the equivalent atoms using a graph neural network-based hydrogenation module; and
[0121] Based on the type of hydrogenation operation corresponding to each non-hydrogen atom of the crystal, the coordinates of each hydrogen atom of the crystal are determined.
[0122] Example 9. The method of Example 8, wherein constructing a geometric relationship diagram of equivalent atoms based on the coordinates and element type of each non-hydrogen atom of the crystal comprises:
[0123] Determine the coordinates and element type of the equivalent atom located within a predetermined distance of each non-hydrogen atom of the crystal by using the symmetry and periodicity of the crystal;
[0124] Calculating the distance between each equivalent atom and every other equivalent atom in the equivalent atoms, wherein the equivalent atoms include non-hydrogen atoms and the determined equivalent atoms; and
[0125] Each pair of equivalent atoms whose distance among the equivalent atoms is less than a distance threshold is taken as a node and connected to an edge, so as to construct a geometric relationship graph of the equivalent atoms.
[0126] Example 10. The method of Example 8, wherein the graph neural network-based hydrogenation module is trained by the following steps:
[0127] receiving a data set comprising a geometric relationship map of equivalent atoms and true values of hydrogenation types of non-hydrogen atoms for each crystal in a plurality of crystals;
[0128] Using the graph neural network-based hydrogenation module, based on the data set, to generate a prediction result of the hydrogenation operation type for each non-hydrogen atom of each crystal in the plurality of crystals;
[0129] calculating a loss function between a predicted result and a true value of the hydrogenation operation type for each non-hydrogen atom of each crystal in the plurality of crystals; and
[0130] Based on the loss function, a parameter updating method is used to update the parameters of the graph neural network-based hydrogenation module to obtain a trained graph neural network-based hydrogenation module.
[0131] Example 11. A system for crystal analysis, comprising means for performing the method as described in any one of Examples 1-10.
[0132] Example 12. A computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the method as described in any one of Examples 1-10.
[0133] Example 13. A computer device comprises a memory and a processor, wherein a computer program capable of being executed on the processor is stored on the memory, wherein when the processor executes the computer program, the method as described in any one of Examples 1-10 is implemented.
[0134] Example 14. A computer program product, comprising a computer program, wherein the computer program, when executed by a processor, implements the method as described in any one of Examples 1-10.
[0135] The present invention may be a system, a method and / or a computer program product. The computer program product may include a computer-readable storage medium carrying computer-readable program instructions for causing a processor to implement various aspects of the present invention.
[0136] Computer readable storage medium can be a tangible device that can hold and store instructions used by an instruction execution device. Computer readable storage medium can be, for example, but not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination thereof. More specific examples (non-exhaustive list) of computer readable storage medium include: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a static random access memory (SRAM), a portable compact disk read-only memory (CD-ROM), a digital versatile disk (DVD), a memory stick, a floppy disk, a mechanical encoding device, for example, a punch card or a convex structure in a groove on which instructions are stored, and any suitable combination thereof. The computer readable storage medium used here is not interpreted as a transient signal itself, such as a radio wave or other freely propagating electromagnetic wave, an electromagnetic wave propagated by a waveguide or other transmission medium (for example, a light pulse by an optical fiber cable), or an electrical signal transmitted by a wire.
[0137] The computer-readable program instructions described herein can be downloaded from a computer-readable storage medium to each computing / processing device, or downloaded to an external computer or external storage device via a network, such as the Internet, a local area network, a wide area network, and / or a wireless network. The network can include copper transmission cables, optical fiber transmissions, wireless transmissions, routers, firewalls, switches, gateway computers, and / or edge servers. The network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards the computer-readable program instructions for storage in the computer-readable storage medium in each computing / processing device.
[0138] The computer program instructions for performing the operation of the present invention may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, state setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages, such as Smalltalk, C++, Python, etc., and conventional procedural programming languages, such as "C" language or similar programming languages. Computer-readable program instructions may be executed entirely on a user's computer, partially on a user's computer, as an independent software package, partially on a user's computer, partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., using an Internet service provider to connect via the Internet). In some embodiments, an electronic circuit, such as a programmable logic circuit, a field programmable gate array (FPGA), or a programmable logic array (PLA), may be personalized by utilizing the state information of the computer-readable program instructions, and the electronic circuit may execute the computer-readable program instructions, thereby realizing various aspects of the present invention.
[0139] Various aspects of the present invention are described herein with reference to the flow charts and / or block diagrams of the methods, devices (systems) and computer program products according to embodiments of the present invention. It should be understood that each box of the flow chart and / or block diagram and the combination of each box in the flow chart and / or block diagram can be implemented by computer-readable program instructions.
[0140] These computer-readable program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, thereby producing a machine, so that when these instructions are executed by the processor of the computer or other programmable data processing device, a device that implements the functions / actions specified in one or more boxes in the flowchart and / or block diagram is generated. These computer-readable program instructions can also be stored in a computer-readable storage medium, and these instructions cause the computer, programmable data processing device, and / or other equipment to work in a specific manner, so that the computer-readable medium storing the instructions includes a manufactured product, which includes instructions for implementing various aspects of the functions / actions specified in one or more boxes in the flowchart and / or block diagram.
[0141] Computer-readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device so that a series of operating steps are performed on the computer, other programmable data processing apparatus, or other device to produce a computer-implemented process, thereby causing the instructions executed on the computer, other programmable data processing apparatus, or other device to implement the functions / actions specified in one or more boxes in the flowchart and / or block diagram.
[0142] The flow charts and block diagrams in the accompanying drawings show the possible architecture, functions and operations of the systems, methods and computer program products according to multiple embodiments of the present invention. In this regard, each box in the flow chart or block diagram can represent a part of a module, a program segment or an instruction, and a part of the module, a program segment or an instruction contains one or more executable instructions for realizing the specified logical function. In some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flow chart, and the combination of the boxes in the block diagram and / or flow chart can be implemented by a dedicated hardware-based system that performs the specified function or action, or can be implemented by a combination of dedicated hardware and computer instructions. It is well known to those skilled in the art that it is equivalent to implement it by hardware, implement it by software, and implement it by combining software and hardware.
[0143] Embodiments of the present invention have been described above, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The selection of terms used herein is intended to best explain the principles of the embodiments, practical applications, or technical improvements in the marketplace, or to enable other persons of ordinary skill in the art to understand the embodiments disclosed herein. The scope of the present invention is defined by the appended claims.
Claims
1. A method for crystal analysis, comprising: Obtaining an electron density map of a crystal, the electron density map being generated by phase resolving diffraction data of the crystal; determining the coordinates of each non-hydrogen atom of the crystal based on the electron density peaks of the electron density map; constructing a geometric relationship diagram of non-hydrogen atoms based on the coordinates of each non-hydrogen atom of the crystal; as well as A non-hydrogen atom identification module based on a graph neural network is used to determine the element type of each non-hydrogen atom of the crystal based on a geometric relationship graph of the non-hydrogen atoms.
2. The method of claim 1, wherein: Constructing a geometric relationship diagram of non-hydrogen atoms based on the coordinates of each non-hydrogen atom of the crystal includes: Calculate the distance from each atom to every other atom among all non-hydrogen atoms; and Each pair of atoms in all non-hydrogen atoms whose distance is less than a distance threshold is taken as a node and connected to an edge to construct a geometric relationship graph of the non-hydrogen atoms.
3. The method of claim 1, wherein: The non-hydrogen atom recognition module based on graph neural network is trained by the following steps: receiving a data set comprising a geometric relationship map of non-hydrogen atoms and true values of elemental types of the non-hydrogen atoms for each of a plurality of crystals; Generate an element type prediction result of each non-hydrogen atom of each crystal in the plurality of crystals based on the data set using the graph neural network-based non-hydrogen atom identification module; calculating a loss function between a predicted result and a true value of the element type of each non-hydrogen atom of each crystal in the plurality of crystals; as well as Based on the loss function, a parameter updating method is used to update the parameters of the graph neural network-based non-hydrogen atom recognition module to obtain a trained graph neural network-based non-hydrogen atom recognition module.
4. The method of claim 1, further comprising: determining a type of hydrogenation operation corresponding to each non-hydrogen atom of the crystal; as well as Based on the type of hydrogenation operation corresponding to each non-hydrogen atom of the crystal, the coordinates of each hydrogen atom of the crystal are determined.
5. The method of claim 4, wherein: Determining the type of hydrogenation operation corresponding to each non-hydrogen atom of the crystal includes: constructing a geometric relationship diagram of equivalent atoms based on the coordinates and element type of each non-hydrogen atom of the crystal, wherein the equivalent atoms include non-hydrogen atoms and equivalent atoms; and A graph neural network-based hydrogenation module is used to determine a type of hydrogenation operation corresponding to each non-hydrogen atom of the crystal based on a geometric relationship graph of the equivalent atoms.
6. The method of claim 5, wherein: Constructing a geometric relationship diagram of equivalent atoms based on the coordinates and element type of each non-hydrogen atom of the crystal includes: Determine the coordinates and element type of the equivalent atom located within a predetermined distance of each non-hydrogen atom of the crystal by using the symmetry and periodicity of the crystal; Calculating the distance between each equivalent atom and every other equivalent atom in the equivalent atoms, wherein the equivalent atoms include non-hydrogen atoms and the determined equivalent atoms; and Each pair of equivalent atoms whose distance among the equivalent atoms is less than a distance threshold is taken as a node and connected to an edge, so as to construct a geometric relationship graph of the equivalent atoms.
7. The method of claim 5, wherein: The graph neural network-based hydrogenation module is trained by the following steps: receiving a data set comprising a geometric relationship map of equivalent atoms and true values of hydrogenation types of non-hydrogen atoms for each crystal in a plurality of crystals; Using the graph neural network-based hydrogenation module, based on the data set, to generate a prediction result of the hydrogenation operation type for each non-hydrogen atom of each crystal in the plurality of crystals; calculating a loss function between a predicted result and a true value of a hydrogenation operation type for each non-hydrogen atom of each crystal in the plurality of crystals; as well as Based on the loss function, a parameter updating method is used to update the parameters of the graph neural network-based hydrogenation module to obtain a trained graph neural network-based hydrogenation module.
8. A method for crystal analysis, comprising: Get the coordinates and element type of each non-hydrogen atom in the crystal; constructing a geometric relationship diagram of equivalent atoms based on the coordinates and element type of each non-hydrogen atom of the crystal, wherein the equivalent atoms include non-hydrogen atoms and equivalent atoms; Determining a type of hydrogenation operation corresponding to each non-hydrogen atom of the crystal based on a geometric relationship graph of the equivalent atoms using a graph neural network-based hydrogenation module; as well as Based on the type of hydrogenation operation corresponding to each non-hydrogen atom of the crystal, the coordinates of each hydrogen atom of the crystal are determined.
9. The method of claim 8, wherein: Constructing a geometric relationship diagram of equivalent atoms based on the coordinates and element type of each non-hydrogen atom of the crystal includes: Determine the coordinates and element type of the equivalent atom located within a predetermined distance of each non-hydrogen atom of the crystal by using the symmetry and periodicity of the crystal; Calculating the distance between each equivalent atom and every other equivalent atom in the equivalent atoms, wherein the equivalent atoms include non-hydrogen atoms and the determined equivalent atoms; and Each pair of equivalent atoms whose distance among the equivalent atoms is less than a distance threshold is taken as a node and connected to an edge, so as to construct a geometric relationship graph of the equivalent atoms.
10. The method of claim 8, wherein: The graph neural network-based hydrogenation module is trained by the following steps: receiving a data set comprising a geometric relationship map of equivalent atoms and true values of hydrogenation types of non-hydrogen atoms for each crystal in a plurality of crystals; Using the graph neural network-based hydrogenation module, based on the data set, to generate a prediction result of the hydrogenation operation type for each non-hydrogen atom of each crystal in the plurality of crystals; calculating a loss function between a predicted result and a true value of a hydrogenation operation type for each non-hydrogen atom of each crystal in the plurality of crystals; as well as Based on the loss function, a parameter updating method is used to update the parameters of the graph neural network-based hydrogenation module to obtain a trained graph neural network-based hydrogenation module.
11. A system for crystal elucidation, comprising means for executing the method according to any one of claims 1 to 10.
12. A computer-readable storage medium having a computer program stored thereon, wherein: When the computer program is executed by a processor, the method according to any one of claims 1 to 10 is implemented.
13. A computer device comprising a memory and a processor, wherein a computer program capable of being run on the processor is stored in the memory, wherein: When the processor executes the computer program, the method according to any one of claims 1 to 10 is implemented.
14. A computer program product comprising a computer program, wherein: When the computer program is executed by a processor, the method according to any one of claims 1 to 10 is implemented.
Citation Information
Patent Citations
Crystalline material analysis device and method, and computer-readable recording medium
CN113223636A
Systems and Methods for Generating Ligand Compounds
US20230317212A1
