High-throughput molecular material screening method, device, electronic device and storage medium

By generating new functional group sets and constructing new molecular structure sets, and using analytical models to screen molecules, the problems of high-throughput molecular screening in the existing technology are solved, and rapid and efficient molecular screening is achieved.

CN114566237BActive Publication Date: 2024-11-12JIHUA LAB
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202210342310.6
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-04-02
Publication Date
2024-11-12
Estimated Expiration
2042-04-02

AI Technical Summary

Technical Problem

The existing high-throughput molecular screening methods are difficult to perform effective screening due to the large amount of direct prediction and calculation of target attributes, and require in-depth domain knowledge to set indirect attributes as screening indicators, resulting in complex processing and difficult to generalize.

Method used

By obtaining molecular component information, a new set of functional groups is generated, and a new set of molecular structures is constructed based on these sets. The preset analysis model is used to screen molecules with target attributes to achieve rapid high-throughput molecular screening, avoiding in-depth domain knowledge rely.

Benefits of technology

The amount of calculation and screening data is reduced, the screening efficiency is improved, the molecules with target attributes can be quickly identified, and the molecular screening process is simplified.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114566237B_ABST
    Figure CN114566237B_ABST
Patent Text Reader

Abstract

The present invention relates to the technical field of molecular screening, and specifically discloses a high-throughput molecular material screening method, device, electronic device and storage medium, wherein the screening method comprises the following steps: obtaining molecular component information; obtaining the existing functional group set of each component according to the molecular component information; obtaining a new functional group set of each component that is different from the existing functional group set according to the molecular component information; generating a new molecular structure set based on the molecular component information according to the existing functional group set and the new functional group set; analyzing the target properties of each molecular structure in the new molecular structure set by using a preset analysis model to screen out a target molecular set; the screening method realizes the directional establishment of a molecular structure database to be screened based on molecular characteristics, can quickly determine the target molecules with the target properties, and completes the screening of high-throughput molecular materials.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of molecular screening technology, and in particular to a high-throughput molecular material screening method, device, electronic equipment and storage medium. Background Art

[0002] Existing high-throughput molecular screening solutions generally establish a method to associate molecular structure with target properties, and then explore specific chemical subspaces to find the molecules (structures) that best meet the target requirements, forming a funnel-like material screening method.

[0003] However, in most scenarios, the direct prediction of the target attribute has the disadvantage of high computational complexity, which makes the screening process difficult to carry out. There are also methods that use indirect attribute quantities related to the target attribute as phased screening indicators to gradually perform predictions or screening, so as to continuously narrow the exploration space and finally obtain the screened molecules. This processing method requires in-depth domain knowledge to identify appropriate indirect attributes as screening conditions, which is difficult to promote.

[0004] There is currently no effective technical solution to the above problems. Summary of the invention

[0005] The purpose of this application is to provide a high-throughput molecular material screening method, device, electronic device and storage medium, to obtain a means of molecular screening with relatively small computational complexity and without the need to set indirect properties based on in-depth domain knowledge.

[0006] In a first aspect, the present application provides a high-throughput molecular material screening method for performing high-throughput molecular screening, the method comprising the following steps:

[0007] Obtain molecular component information;

[0008] Acquire the existing functional group set of each component according to the molecular component information;

[0009] Acquire a new functional group set of each component that is different from the existing functional group set according to the molecular component information;

[0010] Generate a new molecular structure set according to the existing functional group set and the new functional group set based on the molecular component information;

[0011] The target properties of each molecular structure in the new molecular structure set are analyzed using a preset analysis model to screen out a target molecular set.

[0012] The present application discloses a high-throughput molecular material screening method, which obtains a new functional group set of each component that is different from the existing functional group set based on the molecular component information, and constructs a new molecular structure set based on the new functional group set, thereby realizing the establishment of a molecular structure database to be screened based on the molecular characteristics, and can quickly determine the target molecules with target properties, thereby completing the screening of high-throughput molecular materials.

[0013] The high-throughput molecular material screening method, wherein the step of obtaining molecular component information comprises:

[0014] Access existing molecular datasets;

[0015] The molecular combination form in the molecular data set is analyzed to obtain the molecular component information.

[0016] In the high-throughput molecular material screening method of this example, the chemical structure characteristics of the molecular structure can be explained based on prior knowledge in the chemical field or data-driven or existing data characteristics, and the molecular combination forms in the molecular data set can be analyzed based on prior knowledge in the chemical field or data-driven or existing data characteristics to obtain molecular component information.

[0017] The high-throughput molecular material screening method, wherein the preset analysis model is acquired based on the molecular data set training.

[0018] The high-throughput molecular material screening method, wherein the step of generating a new molecular structure set based on the molecular component information according to the existing functional group set and the new functional group set comprises:

[0019] Screening each of the existing functional group sets according to the molecular data set to obtain a corresponding high-weight functional group set;

[0020] Combining the new functional group set and the high-weight functional group set based on the molecular component information to generate a molecular structure set to be tested;

[0021] Analyzing the target attributes of each molecular structure in the set of molecular structures to be tested by using the preset analysis model to screen and obtain a high-potential functional group set in each of the new functional group sets;

[0022] The new molecular structure set is obtained by combining the high potential functional group set and the existing functional group set based on the molecular component information, or combining the high potential functional group set and the existing functional group set and combining the high potential functional group sets of different components.

[0023] In the high-throughput molecular material screening method of this example, the high-potential functional group set and the existing functional group set respectively contain potential unverified functional groups and functional groups in the molecular data set with qualified target attributes. Therefore, the new molecular structure under the combination of the two has not been verified and is likely to produce a molecular structure with qualified target attributes. Similarly, the new molecular structure under the combination of the high-potential functional group set and the high-potential functional group sets of other components has not been verified and is likely to produce a molecular structure with qualified target attributes, and a large number of target molecules will be generated from these unverified new molecular structures.

[0024] In the high-throughput molecular material screening method, the step of using a preset analysis model to analyze the target properties of each molecular structure in the new molecular structure set to screen out the target molecular set includes:

[0025] Analyzing the target attribute of each of the molecular structures in the new molecular structure set by using the preset analysis model;

[0026] The molecular structures whose target attributes are less than a preset threshold in the new molecular structure set are removed to generate the target molecular set.

[0027] In the high-throughput molecular material screening method of this example, the preset threshold is the qualification line of the target attribute in the high-throughput molecular screening, that is, the molecular structures with target attributes above the preset threshold belong to the molecular structures that need to be screened and acquired, and these molecular structures are regarded as target molecules (i.e., high-efficiency molecules). After all target molecules are acquired, a target molecule set is formed to be output as the screening result.

[0028] The high-throughput molecular material screening method further comprises the following steps:

[0029] The target properties of the set of target molecules are verified.

[0030] In the high-throughput molecular material screening method, the step of obtaining a new functional group set of each component that is different from the existing functional group set according to the molecular component information comprises:

[0031] Based on one or two or three of the molecular structure replacement method, the machine learning model generation method and the chemical database extraction method, the new functional group set of each component is obtained according to the molecular component information, and the new functional group set of each component is a non-overlapping set of the existing functional group set of the corresponding component.

[0032] The new functional group set obtained by the above method is a functional group obtained based on the molecular component information. Compared with the chemical subspace established by the existing screening method, it still has the characteristic of small amount of screening data, that is, the useless data is far less than the useless data in the chemical subspace established by the existing technology.

[0033] In a second aspect, the present application also provides a high-throughput molecular material screening device for performing high-throughput molecular screening, the device comprising:

[0034] A first acquisition module is used to acquire molecular component information;

[0035] A second acquisition module is used to acquire the existing functional group set of each component according to the molecular component information;

[0036] A third acquisition module is used to acquire a new functional group set of each component that is different from the existing functional group set according to the molecular component information;

[0037] A combination module, used for generating a new molecular structure set according to the existing functional group set and the new functional group set based on the molecular component information;

[0038] The screening module is used to analyze the target properties of each molecular structure in the new molecular structure set by using a preset analysis model to screen out a target molecular set.

[0039] A high-throughput molecular material screening device of the present application obtains a new functional group set of each component that is different from the existing functional group set based on the molecular component information through a third acquisition module, and constructs a new molecular structure set based on the new functional group set through a combination module, thereby realizing the establishment of a molecular structure database to be screened based on molecular characteristics, and can quickly determine target molecules with target properties, thereby completing the screening of high-throughput molecular materials.

[0040] In a third aspect, the present application further provides an electronic device, comprising a processor and a memory, wherein the memory stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, the steps in the method provided in the first aspect are executed.

[0041] In a fourth aspect, the present application further provides a storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the steps in the method provided in the first aspect are executed.

[0042] As can be seen from the above, the present application provides a high-throughput molecular material screening method, device, electronic device and storage medium, wherein the screening method obtains a new functional group set of each component that is different from the existing functional group set based on the molecular component information, and constructs a new molecular structure set based on the new functional group set. The new molecular structure set belongs to an unverified molecular structure, and its functional group belongs to the functional group in the molecular component information component, so as to realize the establishment of a molecular structure database to be screened based on molecular characteristics. By using a preset analysis model to analyze the target attributes of the molecular structure in the new molecular structure set as the molecular structure database, the target molecules with the target attributes can be quickly determined, and the screening of high-throughput molecular materials can be completed, eliminating the process of using in-depth domain knowledge to set indirect attributes for molecular screening. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 A flow chart of a high-throughput molecular material screening method provided in an embodiment of the present application.

[0044] Figure 2 is the histogram of luminous efficiency distribution.

[0045] Figure 3 Schematic diagram for confirming the peak position of the histogram of luminous efficiency distribution.

[0046] Figure 4 A schematic diagram of the structure of a high-throughput molecular material screening device provided in an embodiment of the present application.

[0047] Figure 5 A more preferred structural schematic diagram of a high-throughput molecular material screening device provided in an embodiment of the present application.

[0048] Figure 6 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application.

[0049] Figure numerals: 201, first acquisition module; 202, second acquisition module; 203, third acquisition module; 204, combination module; 205, screening module; 206, data acquisition module; 207, model training module; 208, verification module; 209, update module; 301, processor; 302, memory; 303, communication bus. DETAILED DESCRIPTION

[0050] The technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all of the embodiments. The components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the application claimed for protection, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present application.

[0051] It should be noted that similar reference numerals and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings. At the same time, in the description of this application, the terms "first", "second", etc. are only used to distinguish the description and cannot be understood as indicating or implying relative importance.

[0052] In most scenarios of high-throughput molecular material screening, since the direct prediction of the target attribute requires too much computational effort or cannot be directly simulated theoretically, the usual practice is to use the indirect attribute quantity related to the target attribute as a phased screening indicator to make the prediction or screening feasible or the computational effort acceptable, and then combine theoretical simulation with machine learning to continuously narrow the exploration space and finally obtain the screened molecules; this method requires in-depth domain knowledge to identify suitable indirect attributes, so it is difficult to promote and has great limitations; there is also a data-driven approach to screening, which generally directly establishes a predictive model from molecular structure to target attribute for direct prediction, but this prediction method requires obtaining a huge database in advance, but usually there is less data in the scenario, resulting in inaccurate predictions by the machine learning model.

[0053] First, please refer to Figure 1-Figure 3 , Figure 1-Figure 3 A high-throughput molecular material screening method in some embodiments of the present application is used for high-throughput molecular screening, and the method comprises the following steps:

[0054] S1. Obtaining molecular component information;

[0055] Specifically, the molecular component information is the combination form of molecular components, reflecting the characteristics of its component combination. For example, toluene can be simply split into two components, methyl and phenyl. Therefore, the molecular component information of toluene can be set to {A, B}, where A and B represent methyl-like and phenyl-like component structures, respectively. The elements in the molecular component information of the molecular structure of multiple components increase accordingly, such as {A, B, C} for three components.

[0056] More specifically, the components included in the molecular component information also characterize the properties of the corresponding components, such as the number of bonds.

[0057] More specifically, since high-throughput molecular screening is based on an existing data set to obtain more molecular structures that meet the target attributes, and since molecular structures with similar combination forms have similar molecular characteristics (physical and chemical properties), molecular structures that meet the target attributes generally have the same combination form, that is, have the same molecular component information, so the molecular component information can serve as the basis for high-throughput molecular screening.

[0058] More specifically, the molecular component information may be obtained by artificial presetting, or may be obtained based on data analysis, model calculation, or chemical prior knowledge.

[0059] More specifically, in the embodiment of the present application, the components in the molecular component information exist in the form of functional groups. Functional groups are atoms or atomic groups that determine the chemical properties of organic compounds and can effectively reflect the physicochemical properties of the molecular structure they constitute.

[0060] S2. Obtaining the existing functional group set of each component according to the molecular component information;

[0061] Specifically, since the components in the molecular component information represent functional groups with similar physicochemical properties, by combining some existing data and based on the characteristic information of the components in the molecular component information, some corresponding existing functional groups in each component can be obtained. These existing functional groups with similar physicochemical properties constitute the existing functional group set of the corresponding components in the molecular component information.

[0062] More specifically, it should be understood that the functional groups in the existing functional group set do not refer to all available functional groups of the corresponding components, but rather functional groups in molecular structures with corresponding target properties that are known or used based on existing experimental data.

[0063] S3, obtaining a new functional group set of each component that is different from the existing functional group set according to the molecular component information;

[0064] Specifically, it can be seen from the above content that each component in the molecular component information has a corresponding existing functional group set, so this step also obtains a new functional group set corresponding to each component.

[0065] More specifically, since the functional groups in the existing functional group set belong to the functional groups that constitute the existing molecular structure with target attributes, it is difficult to obtain a new molecular structure with target attributes in the existing functional group set. Therefore, it is necessary to obtain a new functional group set different from the existing functional group set through step S3 as the basis for obtaining a new molecular structure.

[0066] More specifically, the functional groups in the new functional group set are unverified functional groups to be screened, that is, the molecular structures constructed based on the functional groups in the new functional group set are molecules whose properties have not been verified.

[0067] S4, generating a new molecular structure set according to the existing functional group set and the new functional group set based on the molecular component information;

[0068] Specifically, since the functional groups in the new functional group set are unverified functional groups, the functional groups in the new functional group set and the functional groups in the existing functional group set can form new unverified molecular structures. At the same time, the functional groups in the new functional group set between different components can also form new unverified molecular structures. Since the components have the same characteristics, these unverified molecular structures will have a large number of molecular structures that meet the target attributes.

[0069] More specifically, this step constructs a new molecular structure based on functional groups in a new set of functional groups with corresponding component properties, that is, a new unverified molecular structure library is directionally constructed based on unverified functional groups with the same or similar physicochemical properties, that is, new molecular structures with sufficient number and similar properties are directionally generated based on the molecular composition characteristics for high-throughput molecular screening.

[0070] S5. Analyze the target attributes of each molecular structure in the new molecular structure set using a preset analysis model to screen out the target molecular set.

[0071] Specifically, the preset analysis model is a learning model for analyzing and acquiring target properties of the corresponding molecular structure according to the input molecular structure characteristics.

[0072] Specifically, the target molecule is a molecular structure having qualified target attributes.

[0073] More specifically, the target attribute is the characteristic of the molecular structure obtained by high-throughput molecular screening. For example, if it is necessary to screen and obtain a molecular structure with a sufficiently large luminous efficiency, the target attribute is set to the luminous efficiency. After the molecular structure in the new molecular structure set is input into the analysis model, the analysis model obtains the luminous efficiency of the molecular structure based on the characteristic value analysis of the molecular structure. The molecular structure with a luminous efficiency above a preset threshold belongs to the target molecule, thereby completing the high-throughput molecular screening.

[0074] A high-throughput molecular material screening method according to an embodiment of the present application obtains a new functional group set of each component that is different from the existing functional group set based on molecular component information, and constructs a new molecular structure set based on the new functional group set. The new molecular structure set belongs to unverified molecular structures, and its functional groups belong to the functional groups in the molecular component information components. This realizes the establishment of a molecular structure database to be screened based on molecular characteristics. By using a preset analysis model to analyze the target attributes of the molecular structures in the new molecular structure set as the molecular structure database, the target molecules with the target attributes can be quickly determined, thereby completing the screening of high-throughput molecular materials.

[0075] In some preferred embodiments, step S1 comprises:

[0076] S11, obtaining an existing molecular data set;

[0077] Specifically, the existing molecular data set is a data set consisting of molecular structures with known qualified target attributes.

[0078] More specifically, the molecular dataset includes a plurality of molecular structures that qualify as target attributes.

[0079] More specifically, the molecular dataset can be inputted through a small number of experiments or existing databases or literature.

[0080] S12. Analyze the molecular combination form in the molecular data set to obtain molecular component information.

[0081] Specifically, this step mainly analyzes the molecular characteristics of the molecular structures in the molecular data set to obtain the common characteristics of different molecular structures, thereby confirming the molecular component information.

[0082] More specifically, the combination form of molecular structures (i.e., the component coordination form) is constrained by the molecular structure characteristics and chemical structure properties. Therefore, the combination forms of different molecular structures can be obtained by analyzing the chemical structure properties of the molecular structures in the molecular data set, thereby obtaining molecular component information.

[0083] More specifically, the chemical structure characteristics of the molecular structure can be explained based on prior knowledge in the chemical field or data-driven or existing data characteristics, that is, step S12 can analyze the molecular combination form in the molecular data set based on prior knowledge in the chemical field or data-driven or existing data characteristics to obtain molecular component information.

[0084] In some preferred embodiments, the preset analysis model is obtained based on molecular data set training.

[0085] Specifically, the molecular data set also includes target attribute values ​​corresponding to a plurality of molecular structures that meet the target attribute requirements.

[0086] More specifically, the analysis model is a machine learning model that is end-to-end trained based on the molecular dataset, and is trained according to the molecular structure and corresponding target attribute values ​​in the molecular dataset, so as to obtain an analysis model that can determine the corresponding target attribute values ​​based on the molecular structure.

[0087] More specifically, the analysis model can be trained using common model training methods, such as training based on a graph convolutional neural network (GNN), training based on a sequence model (TRANSFORMER, BERT, etc.), etc.

[0088] More specifically, the target attribute value output by the analysis model is the prediction result.

[0089] In some preferred embodiments, the preset analysis model is a fusion model, that is, the analysis model is obtained by fusing multiple models, and the average result can be calculated based on the analysis results of the multiple models to output as a prediction result, thereby improving the accuracy of the model analysis.

[0090] In some preferred embodiments, step S2 comprises:

[0091] S21. Perform analysis based on chemical prior knowledge and / or based on a molecular data set to obtain optional functional groups of each component in the molecular component information to establish a corresponding existing functional group set.

[0092] Specifically, this step can obtain optional functional groups for each component in the molecular component information based on chemical prior knowledge combined with molecular component information and chemical structure characteristics analysis, and / or disassemble or analyze all molecular structures therein based on the molecular data set to obtain optional functional groups for each component in the molecular component information, and based on these optional functional groups, establish an existing functional group set that includes functional groups in molecular structures that meet the existing target attributes.

[0093] More specifically, analysis based on chemical prior knowledge and molecular data sets can effectively narrow and accurately locate the search scope of existing functional group sets and ensure that the functional groups contained in the existing functional group sets are representative.

[0094] In some preferred embodiments, step S4 comprises:

[0095] S41, screening each existing functional group set according to the molecular data set to obtain a corresponding high-weight functional group set;

[0096] Specifically, the high-weight functional group set includes at least one high-weight functional group, and the high-weight functional group is a representative or relatively important functional group in each high-weight functional group set.

[0097] More specifically, high-weight functional groups are selected from an existing set of functional groups, which are determined by analyzing a molecular dataset to obtain the importance or representativeness of each optional functional group.

[0098] More specifically, high-weight functional groups are selected based on the frequency of occurrence of the functional groups in the molecular data set and / or the average target attribute value in the molecular structure. For example, when the occurrence ratio of a certain functional group in the corresponding component in the molecular data set is greater than a preset ratio, the functional group is extracted and regarded as a high-weight functional group. For example, when the average value of the target attribute of all molecular structures composed of a certain functional group is greater than a preset average value, the functional group is extracted and regarded as a high-weight functional group. By analyzing all functional groups in the molecular data set, the set of high-weight functional groups corresponding to each component can be extracted.

[0099] S42, combining the new functional group set and the high-weight functional group set based on the molecular component information to generate a molecular structure set to be tested;

[0100] Specifically, since the functional groups in the new functional group sets have not been verified, that is, the molecular structures that combine these functional groups with existing or non-existing functional groups based on the molecular component information all belong to unverified target attributes, multiple molecular structures with qualified target attributes can be quickly explored based on these new functional group sets.

[0101] More specifically, since there are a large number of functional groups in the new functional group set and the chemical structure characteristics corresponding to the molecular structure searched based on these functional groups are unknown, in order to reduce the amount of subsequent screening data, it is necessary to screen the functional groups in the new functional group set to obtain some functional groups that have certain potential to constitute molecular structures with qualified target attributes. Therefore, step S42 combines the functional groups of the new functional group set and the high-weight functional group set based on the molecular component information to form a molecular structure to be tested, so as to analyze the molecular structure set to be tested composed of these molecular structures to be tested and screen out the functional groups in the new functional group set with insufficient potential to reduce the amount of subsequent data analysis.

[0102] More specifically, in the process of generating the molecular structure to be tested, it is necessary to combine the components according to the component characteristics of the molecular component information. For example, for the molecular component information of {A, B}, the functional groups in the new functional group set corresponding to component A and the functional groups in the high-weight functional group set corresponding to component B are combined, and the functional groups in the new functional group set corresponding to component B and the functional groups in the high-weight functional group set corresponding to component A are combined, so as to obtain the molecular structure to be tested and constitute the molecular structure set to be tested.

[0103] More specifically, since high-weight functional groups are representative, the molecular structure to be tested formed by combining the functional groups in the new functional group set with the functional groups in the high-weight functional groups can serve as a preliminary verification of whether the functional groups in the new functional group set have the basic structure to constitute a molecular structure with qualified target attributes.

[0104] S43, using a preset analysis model to analyze the target attributes of each molecular structure in the set of molecular structures to be tested to screen and obtain a high-potential functional group set in each new functional group set;

[0105] Specifically, this step mainly analyzes the set of molecular structures to be tested that is established based on representative high-weight functional groups, and screens out the functional groups of the new functional group set corresponding to the molecular structures to be tested that have lower target attributes or do not meet expectations, so as to narrow the set of new functional groups and obtain a high-potential functional group set, that is, retain some functional groups that have the potential to constitute a molecular structure with qualified target attributes.

[0106] More specifically, this step can be implemented in at least one of the following two ways: First, analyzing the target attributes of all molecular structures in the set of molecular structures to be tested through an analysis model, screening out the molecular structures to be tested whose target attributes are lower than expected, and extracting the functional groups in the corresponding new functional group set from the remaining molecular structures after the screening process to form a high-potential functional group set composed of functional groups with potential for the corresponding component; Second, performing a statistical target attribute analysis on all molecular structures to be tested composed of the functional groups in each new functional group set, and when the statistical target attribute meets expectations, the functional group is regarded as a high-potential functional group, and after analyzing the functional groups in all new functional group sets, extracting the functional groups in the new functional group set corresponding to the molecular structures to be tested whose target attributes are greater than or equal to the expected value to form a high-potential functional group set composed of functional groups with potential for the corresponding component.

[0107] More specifically, step S43 of the high-throughput analytical material screening method of the embodiment of the present application preferably adopts the second method to obtain a high-potential functional group set.

[0108] More specifically, the statistical target attribute analysis is preferably performed using a histogram, that is, judging whether the corresponding functional group is a high-potential functional group based on the distribution or form of the target attributes of different molecular structures on the histogram.

[0109] More specifically, before executing step S43, the molecular structures to be tested in the set of molecular structures to be tested can be screened out based on indicators such as chemical feasibility, molecular toxicity, solubility, etc., so as to further reduce the amount of data analysis, improve the efficiency of obtaining high-potential functional group sets, and thereby improve the efficiency of molecular screening.

[0110] S44. Combining a high-potential functional group set with an existing functional group set based on molecular component information, or combining a high-potential functional group set with an existing functional group set and combining high-potential functional group sets of different components to obtain a new molecular structure set.

[0111] Specifically, the high-potential functional group set and the existing functional group set respectively contain potential unverified functional groups and functional groups in the molecular data set with qualified target attributes. Therefore, the new molecular structure under the combination of the two has not been verified and is likely to produce a molecular structure with qualified target attributes. Similarly, the new molecular structure under the combination of the high-potential functional group set and the high-potential functional group sets of other components has not been verified and is likely to produce a molecular structure with qualified target attributes. A large number of target molecules will be generated from these unverified new molecular structures.

[0112] In other preferred embodiments, in order to simplify the calculation logic, step S4 can skip the process of obtaining the high-weight functional group set, and only use the new functional group set and the existing functional group set to calculate and obtain the high-potential functional group set. In this case, step S4 includes:

[0113] S41', combining the new functional group set and the existing functional group set based on the molecular component information to generate a molecular structure set to be tested;

[0114] S42', using a preset analysis model to analyze the target attributes of each molecular structure in the set of molecular structures to be tested to screen and obtain a high-potential functional group set in each new functional group set;

[0115] S43', based on the molecular component information, combining the high potential functional group set with the existing functional group set, or combining the high potential functional group set with the existing functional group set and combining the high potential functional group set with the high potential functional group set of the remaining components to obtain a new molecular structure set.

[0116] Specifically, this processing method skips the screening process of high-weight functional group sets and directly uses the new functional group set and the existing functional group set to generate the molecular structure set to be tested. It can simplify the calculation logic, relatively increase the data volume of the molecular structure set to be tested, and increase the amount of data analysis to a certain extent, but can ensure that the obtained high-potential functional group set is more accurate.

[0117] In the high-throughput molecular material screening method of the embodiment of the present application, it is preferred to adopt steps S41-S44 to obtain a new molecular structure set to gradually reduce the amount of data analysis for molecular screening, thereby improving the overall screening efficiency.

[0118] In some preferred embodiments, step S5 comprises:

[0119] S51, analyzing the target attribute of each molecular structure in the new molecular structure set by using a preset analysis model;

[0120] Specifically, before executing step S51, the molecular structures in the new molecular structure set can be screened out based on indicators such as chemical feasibility, molecular toxicity, solubility, etc., so as to reduce the amount of data analysis in step S51 and thereby improve the efficiency of molecular screening.

[0121] S52, removing molecular structures whose target attributes are less than a preset threshold in the new molecular structure set to generate a target molecular set.

[0122] Specifically, the preset threshold is the qualified line of the target attribute in high-throughput molecular screening, that is, the molecular structures with target attributes above the preset threshold belong to the molecular structures that need to be screened and acquired, and these molecular structures are regarded as target molecules (i.e., high-efficiency molecules). After all target molecules are acquired, a target molecule set is formed to be output as the screening result.

[0123] In some preferred embodiments, step S5 further includes:

[0124] S53, acquiring a target functional group combination set according to the target molecule set, where the target functional group combination is a combination type of functional groups in the molecular structure of the target molecule set;

[0125] S54, expanding the molecular combination form according to the functional group combination set to obtain an expanded target molecular set;

[0126] Specifically, after obtaining the target molecules that meet the expectations, it is possible to expand the components in different proportions according to the functional group combination of the target molecules to obtain more target molecules. For example, for the target molecules of {A, B} components, if an A1 functional group can connect two B1 functional groups, then based on the new molecular structure of A1-B1, it is possible to expand the target molecules of B1-A1-B1, thereby obtaining more target molecules that meet the expectations.

[0127] S55. After chemical synthesis screening, the expanded target molecule set is added to the target molecule set.

[0128] Specifically, the expanded target molecules need to be verified whether they can be obtained through chemical synthesis. After screening out the expanded target molecules that cannot be obtained by synthesis, the remaining expanded target molecules are regarded as additional target molecules and integrated into the target molecules as the screening result output.

[0129] In some preferred embodiments, the high-throughput molecular material screening method of the present application embodiment further comprises the following steps:

[0130] S6. Verify the target properties of the target molecule set.

[0131] Specifically, steps S1-S5 mainly perform calculation and analysis of target properties through the analysis model to determine the target molecule set. The analysis model may have certain analysis errors and affect the final output results. Therefore, it is necessary to add step S6 to verify whether the molecular structure in the target molecule set is exactly as expected, that is, whether it has qualified target properties.

[0132] More specifically, this step is mainly to verify the target properties of the target molecule set by experimental testing, and the molecular structures that do not meet the target properties need to be removed from the target molecule set.

[0133] In some preferred embodiments, the high-throughput molecular material screening method of the present application embodiment further comprises the following steps:

[0134] S7. Update the molecular dataset according to the target molecule set, or update the molecular dataset and the analysis model.

[0135] Specifically, after obtaining a new target molecule set, the molecular structures and corresponding target attributes of the target molecules in the target molecule set can be added to the molecular data set to update the molecular data set, so as to serve as the data basis for the next screening.

[0136] More specifically, after the molecular dataset is updated, the updated molecular dataset may be used to update the analysis model or retrain the corresponding analysis model to serve as the analysis model used for the next screening.

[0137] In some preferred embodiments, step S3 comprises:

[0138] S31. Based on one or two or three of the methods selected from the group consisting of molecular structure replacement, machine learning model generation, and chemical database extraction, a new functional group set for each component is obtained according to the molecular component information. The new functional group set for each component is a non-overlapping set of the existing functional group set of the corresponding component, that is, all elements contained in the new functional group set are not in the existing functional group set of the corresponding component.

[0139] Specifically, in order to obtain as many new functional groups as possible to expand the search scope of the molecular structure, in the embodiments of the present application, it is preferred to use all of the above methods to obtain a new set of functional groups for each component.

[0140] Specifically, the process of obtaining a new functional group set based on molecular structure replacement, machine learning model generation and chemical database extraction is as follows: replacing part of the structure of the functional groups in the existing functional group set with other similar structures, generating new functional group structures based on the machine learning model generation, and extracting functional group structures with similar properties based on the chemical database to obtain a new functional group set.

[0141] More specifically, the new functional group set obtained by the above method is the functional group obtained based on the molecular component information. Compared with the chemical subspace established by the existing screening method, the amount of screening data is still small, that is, the useless data is far less than the useless data in the chemical subspace established by the existing technology.

[0142] Example 1

[0143] In order to more clearly illustrate the implementation process of the high-throughput molecular material screening method of the embodiment of the present application, the high-throughput molecular screening process of TADF (Thermally activated delayed fluorescence) molecules is described below as an example:

[0144] For TADF molecular screening, the target property of interest is luminescence efficiency, which is defined in the interval [0,1], and the higher the luminescence efficiency, the better.

[0145] Based on existing literature data and self-experimental data, a molecular data set D = {M, L} is obtained, where M is the molecular structure and L is the corresponding luminous efficiency. The molecular data set contains about 1000 M.

[0146] Analysis of the TADF molecular data set shows that the characteristics of TADF molecules are usually composed of two parts, namely the acceptor and the donor, which are marked as A and B respectively, thereby determining the molecular component information of TADF as {A, B}.

[0147] By analyzing the molecular data set D and obtaining the optional functional groups of the receptor and donor, it can be seen that there are about 300 functional groups for the receptor and about 150 functional groups for the donor, which constitute the corresponding existing functional group sets, respectively denoted as A i and B i .

[0148] In order to screen representative high-weight functional groups, the luminescence efficiency threshold was set to 0.8, and the molecular structures in some molecular data sets were screened out. The occurrence counts of all functional groups in the remaining molecular structures were obtained, and the 10 functional groups with the highest occurrence counts in the acceptor and donor were extracted to establish the corresponding high-weight functional group sets as backup, which were recorded as A and B. h and B h .

[0149] The molecular data set is used to train an analysis model for future use, and the analysis model can predict the corresponding luminescence efficiency based on the molecular structure.

[0150] Based on one, two or three of the molecular structure replacement method, the machine learning model generation method and the chemical database extraction method, functional groups different from the existing functional group set are correspondingly obtained, and these functional groups are combined to form new functional group sets of receptors and donors, respectively, which are recorded as A new , B new .

[0151] The functional groups of the new functional group set are combined with the functional groups of the high-weight functional group set to obtain the molecular structure set to be tested, and then the luminescence efficiency of the molecular structure of the molecular structure set to be tested is analyzed using the analysis model, and then the corresponding high-potential functional group set is obtained from the new functional group set of the acceptor and the donor, which are respectively recorded as A pot , B pot .

[0152] Specifically, a functional group a in a new functional group set of a receptor new,1 Take a high potential functional group as an example: new,1 With B h All high-weight functional groups in are combined to generate corresponding molecular structures, which are recorded as {a new,1 -b h}, where b h For B h The high-weight functional groups in the new,1 -b h}, directly predict and obtain all a new,1 With B h The luminescence efficiency of the functional group combination molecules in the Figure 2 The histogram shown in the figure has a horizontal coordinate that divides the luminous efficiency [0,1] into 10 intervals, and a vertical ... new,1 With B h Distribution of luminous efficiency of combined molecular structures.

[0153] More specifically, although B h There are only 10 elements, but since functional groups generally have multiple attachment sites, a new,1 With B h The molecular structure of the combination is more than 10. If a new,1 and b h has 5 connection sites, then {a new,1 -b h It has 5*5*1*10, or 250, molecular structures, and can screen out inappropriate molecular structures based on indicators such as chemical feasibility, molecular toxicity, and solubility.

[0154] By analyzing the functional groups in all new functional group sets, a histogram equal to the number of functional groups can be obtained. High-potential functional groups can be screened out according to the distribution of the luminous efficiency of the histogram obtained for each functional group. The screening process can set a threshold judgment or determine the high-potential functional groups according to the peak position of the histogram, such as Figure 3 The figure shows that the high potential functional group is determined by the histogram peak position, that is, the abscissa of the corresponding histogram peak position is obtained by peak position calculation, and whether the abscissa is greater than the preset threshold is determined to determine whether the corresponding functional group is a high potential functional group. After completing the analysis and judgment of all histograms, A can be obtained. pot , B pot .

[0155] After obtaining the high potential functional group set, a new molecular structure set is obtained based on the combination of the high potential functional group set. In this embodiment, a new molecular structure set is obtained by combining the high potential functional group set with the existing functional group set and by combining the high potential functional group set, that is, by A pot With B i Combination, A i With B pot Combination, A pot With B pot Combination, and screening out inappropriate molecular structures based on indicators such as chemical feasibility, molecular toxicity, solubility, etc. to generate a new molecular structure collection.

[0156] After obtaining the new molecular structure set, the molecular structures in the new molecular structure set are analyzed again through the analysis model to obtain molecular structures whose luminous efficiency meets the expected requirements, so as to form a target molecular set; then, according to the functional group combination form in the target molecular set, the same receptor or donor is expanded based on the connection site to further obtain an expanded molecular set, and the analysis model is used to analyze and add qualified expanded molecular structures to the target molecular set to update the target molecular set. If the number of target molecular sets is large, the target molecular set can be narrowed based on theoretical simulation and other methods.

[0157] Finally, the target molecule set is verified through experimental verification and TADF molecular structures with high luminescence efficiency are screened out. These molecular structures are used to update the molecular data set and analysis model for the next high-throughput molecular screening.

[0158] Second, please refer to Figure 4 and Figure 5 , Figure 4 and Figure 5 A high-throughput molecular material screening device provided in some embodiments of the present application is used for high-throughput molecular screening, and the device comprises:

[0159] The first acquisition module 201 is used to acquire molecular component information;

[0160] A second acquisition module 202 is used to acquire the existing functional group set of each component according to the molecular component information;

[0161] The third acquisition module 203 is used to acquire a new functional group set of each component that is different from the existing functional group set according to the molecular component information;

[0162] A combination module 204, for generating a new molecular structure set according to the existing functional group set and the new functional group set based on the molecular component information;

[0163] The screening module 205 is used to analyze the target attribute of each molecular structure in the new molecular structure set by using a preset analysis model to screen out the target molecular set.

[0164] A high-throughput molecular material screening device according to an embodiment of the present application obtains a new functional group set of each component that is different from the existing functional group set based on molecular component information through a third acquisition module 203, and constructs a new molecular structure set based on the new functional group set through a combination module 204, thereby establishing a molecular structure database to be screened based on molecular characteristics. By using a preset analysis model to analyze the target attributes of the molecular structures in the new molecular structure set serving as the molecular structure database, the target molecules with the target attributes can be quickly determined, thereby completing the screening of high-throughput molecular materials.

[0165] In some preferred embodiments, the high-throughput molecular material screening device further comprises:

[0166] The data acquisition module 206 is used to acquire an existing molecular data set. The first acquisition module 201 analyzes the molecular combination form in the molecular data set acquired by the data acquisition module 206 to acquire molecular component information.

[0167] In some preferred embodiments, the high-throughput molecular material screening device further comprises:

[0168] The model training module 207 is used to obtain a preset analysis model based on the molecular data set training.

[0169] In some preferred embodiments, the high-throughput molecular material screening device further comprises:

[0170] The verification module 208 is used to verify the target properties of the target molecule set.

[0171] In some preferred embodiments, the high-throughput molecular material screening device further comprises:

[0172] The updating module 209 is used to update the molecular data set according to the target molecule set, or to update the molecular data set and the analysis model.

[0173] In some preferred embodiments, the high-throughput molecular material screening device of the embodiments of the present application is used to execute the high-throughput molecular material screening method provided in the first aspect above.

[0174] Third, please refer to Figure 6 , Figure 6 This is a structural diagram of an electronic device provided in an embodiment of the present application. The present application provides an electronic device, including: a processor 301 and a memory 302. The processor 301 and the memory 302 are interconnected and communicate with each other through a communication bus 303 and / or other forms of connection mechanisms (not shown). The memory 302 stores a computer program executable by the processor 301. When the computing device is running, the processor 301 executes the computer program to execute the method in any optional implementation of the above-mentioned embodiment.

[0175] In a fourth aspect, an embodiment of the present application provides a storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the method in any optional implementation of the above embodiment is executed. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, disk or optical disk.

[0176] In summary, the embodiments of the present application provide a high-throughput molecular material screening method, device, electronic device and storage medium, wherein the screening method obtains a new functional group set of each component that is different from the existing functional group set based on the molecular component information, and constructs a new molecular structure set based on the new functional group set. The new molecular structure set belongs to an unverified molecular structure, and its functional group belongs to the functional group in the molecular component information component, so as to realize the establishment of a molecular structure database to be screened based on molecular characteristics. By using a preset analysis model to analyze the target attributes of the molecular structure in the new molecular structure set as the molecular structure database, the target molecules with the target attributes can be quickly determined, and the screening of high-throughput molecular materials can be completed, eliminating the process of using in-depth domain knowledge to set indirect attributes for molecular screening.

[0177] In the embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interfaces, and the indirect coupling or communication connection of the devices or units can be electrical, mechanical or other forms.

[0178] In addition, the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0179] Furthermore, the functional modules in the various embodiments of the present application may be integrated together to form an independent part, or each module may exist separately, or two or more modules may be integrated to form an independent part.

[0180] In this document, relational terms such as first and second, etc. are used merely to distinguish one entity or operation from another entity or operation, but do not necessarily require or imply any such actual relationship or order between these entities or operations.

[0181] The above description is only an embodiment of the present application and is not intended to limit the protection scope of the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A high-throughput molecular material screening method for high-throughput molecular screening, characterized in that: The method comprises the following steps: Obtain molecular component information; Acquire the existing functional group set of each component according to the molecular component information; Acquire a new functional group set of each component that is different from the existing functional group set according to the molecular component information, wherein the functional groups in the new functional group set are unverified functional groups to be screened; Generate a new molecular structure set according to the existing functional group set and the new functional group set based on the molecular component information, wherein the new molecular structure set belongs to an unverified molecular structure; Analyzing the target properties of each molecular structure in the new molecular structure set by using a preset analysis model to screen out a target molecular set; The step of obtaining molecular component information comprises: Access existing molecular datasets; Analyzing the molecular combination form in the molecular data set to obtain the molecular component information; The step of generating a new molecular structure set based on the molecular component information according to the existing functional group set and the new functional group set comprises: Screening each of the existing functional group sets according to the molecular data set to obtain a corresponding high-weight functional group set; Combining the new functional group set and the high-weight functional group set based on the molecular component information to generate a molecular structure set to be tested; Analyzing the target attributes of each molecular structure in the set of molecular structures to be tested by using the preset analysis model to screen and obtain a high-potential functional group set in each of the new functional group sets; The new molecular structure set is obtained by combining the high potential functional group set and the existing functional group set based on the molecular component information, or, the new molecular structure set is obtained by combining the high potential functional group set and the existing functional group set and combining the high potential functional group sets of different components based on the molecular component information.

2. The high-throughput molecular material screening method according to claim 1, characterized in that: The preset analysis model is acquired based on the molecular data set training.

3. The high-throughput molecular material screening method according to claim 1, characterized in that: The step of using a preset analysis model to analyze the target attributes of each molecular structure in the new molecular structure set to screen out the target molecular set includes: Analyzing the target attribute of each of the molecular structures in the new molecular structure set by using the preset analysis model; The molecular structures whose target attributes are less than a preset threshold in the new molecular structure set are removed to generate the target molecular set.

4. The high-throughput molecular material screening method according to claim 1, characterized in that: The method further comprises the following steps: The target properties of the set of target molecules are verified.

5. The high-throughput molecular material screening method according to claim 1, characterized in that: The step of obtaining a new functional group set of each component that is different from the existing functional group set according to the molecular component information comprises: Based on one or two or three of the molecular structure replacement method, the machine learning model generation method and the chemical database extraction method, the new functional group set of each component is obtained according to the molecular component information, and the new functional group set of each component is a non-overlapping set of the existing functional group set of the corresponding component.

6. A high-throughput molecular material screening device for high-throughput molecular screening, characterized in that: The device comprises: A first acquisition module is used to acquire molecular component information; A second acquisition module is used to acquire the existing functional group set of each component according to the molecular component information; A third acquisition module is used to acquire a new functional group set of each component that is different from the existing functional group set according to the molecular component information, wherein the functional groups in the new functional group set are unverified functional groups to be screened; A combination module, used for generating a new molecular structure set according to the existing functional group set and the new functional group set based on the molecular component information, wherein the new molecular structure set belongs to an unverified molecular structure; A screening module, used to analyze the target properties of each molecular structure in the new molecular structure set by using a preset analysis model to screen out a target molecular set; The step of obtaining molecular component information comprises: Access existing molecular datasets; Analyzing the molecular combination form in the molecular data set to obtain the molecular component information; The step of generating a new molecular structure set based on the molecular component information according to the existing functional group set and the new functional group set comprises: Screening each of the existing functional group sets according to the molecular data set to obtain a corresponding high-weight functional group set; Combining the new functional group set and the high-weight functional group set based on the molecular component information to generate a molecular structure set to be tested; Analyzing the target attributes of each molecular structure in the set of molecular structures to be tested by using the preset analysis model to screen and obtain a high-potential functional group set in each of the new functional group sets; The new molecular structure set is obtained by combining the high potential functional group set and the existing functional group set based on the molecular component information, or, the new molecular structure set is obtained by combining the high potential functional group set and the existing functional group set and combining the high potential functional group sets of different components based on the molecular component information.

7. An electronic device, characterized in that: The method comprises a processor and a memory, wherein the memory stores computer-readable instructions. When the computer-readable instructions are executed by the processor, the steps in the method according to any one of claims 1 to 5 are executed.

8. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are executed.