High-throughput screening method, device, equipment, storage medium and product for molecules

Abstract

The present application discloses a method, device, equipment, storage medium and product for high-throughput screening of molecules, which relates to the technical field of organic materials, and discloses a method for high-throughput screening of molecules, including: obtaining molecular orbital descriptors of the molecules to be screened; inputting the molecular orbital descriptors into a preset prediction model to predict the energy level difference of each molecule to be screened, the preset prediction model is obtained by training a basic machine learning model using the molecular orbital descriptors of each sample D-A-D molecule as input and the energy level difference of each sample D-A-D molecule as output; based on the ordering of the energy level difference from large to small, target molecules that meet the preset ranking are screened from each molecule to be screened. By obtaining the molecular orbital descriptors of each molecule to be screened, the hybridized part in the D-A-D molecule is simplified, the difficulty of screening molecules according to the hybridized part is reduced, and the accuracy of screening is improved by using the preset prediction model, thereby improving the efficiency of high-throughput screening of molecules.

Description

Technical Field

[0001] The present application relates to the technical field of organic materials, and in particular to methods, devices, equipment, storage media and products for high-throughput screening of molecules. Background Art

[0002] The DAD (Donor-Acceptor-Donnor) system has attracted much attention due to its characteristics such as local hybridization and charge transfer, which improve the utilization rate of excitons. Therefore, it has become an important research direction for the next generation of OLED (Organic Light-Emitting Diode) materials.

[0003] The two donor units of the DAD system material can collect singlet and triplet excitons simultaneously through direct electronic interaction with the acceptor unit. In order to prevent exciton quenching caused by long-term operation, it is necessary to select DAD molecules that satisfy the small enough energy difference (ΔE) between the singlet state Sm and the triplet state Tn. ST ), and a large energy gap (ΔE) between the triplet state (T2 or more commonly Tn) and T1 (or more commonly Tn-1). TT ) molecules and inhibit the internal conversion decay of Tn, preventing the accumulation of triplet excitons within the molecule. However, due to the local hybridization inside DAD, the high-throughput screening efficiency of molecules is low.

[0004] The above contents are only used to assist in understanding the technical solution of the present application and do not constitute an admission that the above contents are prior art. Summary of the invention

[0005] The main purpose of this application is to provide a high-throughput screening method for molecules, aiming to solve the technical problem of low efficiency in high-throughput screening of molecules.

[0006] To achieve the above objectives, the present application proposes a high-throughput screening method for molecules, the method comprising:

[0007] Obtaining molecular orbital descriptors of the molecules to be screened;

[0008] Inputting the molecular orbital descriptor into a preset prediction model to predict the energy level difference of each molecule to be screened, wherein the preset prediction model is obtained by training a basic machine learning model using the molecular orbital descriptor of each sample DAD molecule as input and the energy level difference of each sample DAD molecule as output;

[0009] Based on the order of the energy level differences from large to small, target molecules that meet the preset ranking are screened from the molecules to be screened.

[0010] In one embodiment, the step of obtaining the molecular orbital descriptor of the molecule to be screened includes:

[0011] Identify the donors and acceptors that make up the molecule to be screened;

[0012] Based on a preset DFT theory, determining the highest orbital descriptor, the lowest orbital descriptor, the occupied orbital descriptor and the unoccupied orbital descriptor of the donor and the acceptor respectively;

[0013] Among them, the highest orbital descriptor is the descriptor of the highest occupied molecular orbital, the lowest orbital descriptor is the descriptor of the lowest unoccupied molecular orbital, the occupied orbital descriptor is the descriptor of an occupied molecular orbital at a preset energy level, and the unoccupied orbital descriptor is the descriptor of an unoccupied molecular orbital at a preset energy level.

[0014] In one embodiment, the step of identifying the donor and the acceptor constituting the molecule to be screened further comprises:

[0015] Retrieving molecular fragments in historical data, and counting the occurrence frequency of each molecular fragment in the historical data;

[0016] Selecting a donor molecule fragment, an acceptor molecule fragment, and a substituent fragment from the molecule fragments based on the occurrence frequency;

[0017] A DAD molecule library storing the molecules to be screened is created based on the donor molecule fragment, the acceptor molecule fragment and the substituent fragment.

[0018] In one embodiment, the step of obtaining the molecular orbital descriptor of the molecule to be screened further includes:

[0019] Selecting a preset number of sample DAD molecules from a DAD molecule library storing the molecules to be screened;

[0020] determining the energy level difference of the sample DAD molecules;

[0021] Based on the energy level difference, selecting a basic machine learning model with the strongest learning ability from a plurality of machine learning models, wherein the basic machine learning models include at least one;

[0022] Based on the molecular orbital descriptors of the sample DAD molecules and the corresponding energy level differences, the basic machine learning model is iteratively trained to obtain a preset prediction model.

[0023] In one embodiment, the step of iteratively training the basic machine learning model based on the molecular orbital descriptors of the sample DAD molecules and the corresponding energy level differences to obtain a preset prediction model comprises:

[0024] Based on a preset random search method, the basic machine learning model is preliminarily iteratively trained according to the molecular orbital descriptors of the sample DAD molecules and the corresponding energy level differences to obtain a preliminary iterative training model;

[0025] Based on a preset grid search method, the preliminary iterative training model is finally iteratively trained according to the molecular orbital descriptors of the sample DAD molecules and the corresponding energy level differences to obtain a preset prediction model.

[0026] In one embodiment, if there are multiple preset prediction models, the step of selecting target molecules that meet the preset ranking from the molecules to be screened based on the order of the energy level differences from large to small includes:

[0027] The energy level differences predicted by the different preset prediction models are sorted from large to small to obtain corresponding property sorting;

[0028] From the rankings of the properties, extract the molecules to be analyzed that are ranked before a preset ranking;

[0029] Sort the molecules to be analyzed in the property rankings from large to small according to the energy level differences to obtain a comprehensive ranking;

[0030] Based on the comprehensive ranking, the target molecules with the preset ranking are screened from the molecules to be analyzed.

[0031] In addition, to achieve the above-mentioned purpose, the present application also proposes a high-throughput screening device for molecules, the high-throughput screening device for molecules comprising:

[0032] An acquisition module, used for acquiring molecular orbital descriptors of molecules to be screened;

[0033] A prediction module, used to input the molecular orbital descriptor into a preset prediction model to predict the energy level difference of each molecule to be screened, wherein the preset prediction model is obtained by training a basic machine learning model using the molecular orbital descriptor of each sample DAD molecule as input and the energy level difference of each sample DAD molecule as output;

[0034] The screening module is used to screen the target molecules that meet the preset ranking from the molecules to be screened based on the order of the energy level differences from large to small.

[0035] In addition, to achieve the above-mentioned objectives, the present application also proposes a high-throughput screening device for molecules, which includes: a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the computer program is configured to implement the steps of the high-throughput screening method for molecules as described above.

[0036] In addition, to achieve the above objectives, the present application also proposes a storage medium, which is a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the high-throughput screening method for molecules as described above are implemented.

[0037] In addition, to achieve the above objectives, the present application also provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, it implements the steps of the high-throughput screening method for molecules as described above.

[0038] One or more technical solutions proposed in this application have at least the following technical effects:

[0039] Since the molecular orbital descriptors can be used to characterize the behavior of each electron in a molecule, the molecular orbital descriptors of the molecules to be screened are obtained, and then the preset prediction model is used to predict the energy level difference of each molecule to be screened. With the help of the accuracy of the model prediction, the molecular orbital descriptors are used as the screening basis to predict the energy level difference of each molecule to be screened. The preset prediction model is obtained by training a basic machine learning model with the molecular orbital descriptors of the sample DAD molecules as input and the energy level difference of each sample DAD molecule as output, thereby ensuring the accuracy of predicting the energy level difference of each molecule to be screened. Then, according to the size of the energy level difference, the target molecules that meet the preset ranking are screened from the molecules to be screened to achieve high-throughput screening of molecules. That is, by obtaining the molecular orbital descriptors of each molecule to be screened, the hybrid part in the DAD molecule is simplified, the difficulty of screening molecules according to the hybrid part is reduced, and the preset prediction model is used to improve the accuracy of screening, thereby improving the efficiency of high-throughput screening of molecules. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0041] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0042] Figure 1 A schematic diagram of the process of high-throughput screening of molecules according to the present application is provided in Example 1;

[0043] Figure 2 A schematic diagram of the process provided for Example 2 of the high-throughput screening method for molecules of the present application;

[0044] Figure 3 Schematic diagram of donor examples provided for the high-throughput screening method of the molecules of the present application;

[0045] Figure 4 Schematic diagram of an example of a receptor provided for the high throughput screening method of the molecules of the present application;

[0046] Figure 5 Schematic diagram of examples of substituents provided for the high-throughput screening method of the molecules of the present application;

[0047] Figure 6 A schematic diagram of the process provided for the high-throughput screening method of the molecules of the present application in Example 3;

[0048] Figure 7 This is a schematic diagram of the module structure of a high-throughput screening device for molecules according to an embodiment of the present application;

[0049] Figure 8 Schematic diagram of the equipment structure of the hardware operating environment involved in the high-throughput screening method for molecules in the embodiments of the present application.

[0050] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0051] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of the present application and are not used to limit the present application.

[0052] In order to better understand the technical solution of the present application, a detailed description will be given below in conjunction with the accompanying drawings and specific implementation methods.

[0053] The main solution of the embodiment of the present application is: the molecular screening platform obtains the molecular orbital descriptors of the molecules to be screened; the molecular orbital descriptors are input into a preset prediction model to predict the energy level difference of each molecule to be screened, and the preset prediction model is obtained by training a basic machine learning model with the molecular orbital descriptors of each sample DAD molecule as input and the energy level difference of each sample DAD molecule as output; based on the order of the energy level differences from large to small, target molecules that meet the preset ranking are screened from each molecule to be screened.

[0054] In this embodiment, for the convenience of description, the molecular screening platform is used as the execution body for explanation below.

[0055] Since the two donor units of the DAD system material can collect singlet and triplet excitons simultaneously through direct electronic interaction with the acceptor unit, in order to prevent exciton quenching caused by long-term operation, it is necessary to screen DAD molecules that satisfy the small enough energy difference (ΔE) between the singlet state Sm and the triplet state Tn. ST), and a large energy gap (ΔE) between the triplet state (T2 or more commonly Tn) and T1 (or more commonly Tn-1). TT ) molecules and inhibit the internal conversion decay of Tn, preventing the accumulation of triplet excitons within the molecule. However, due to the local hybridization inside DAD, the high-throughput screening efficiency of molecules is low.

[0056] The present application provides a solution. Since the molecular orbital descriptors can be used to characterize the behavior of each electron in a molecule, the molecular orbital descriptors of the molecules to be screened are obtained, and then the energy level differences of each molecule to be screened are predicted using a preset prediction model. With the help of the accuracy of the model prediction, the molecular orbital descriptors are used as the screening basis to predict the energy level differences of each molecule to be screened. The preset prediction model is obtained by training a basic machine learning model with the molecular orbital descriptors of the sample DAD molecules as input and the energy level differences of each sample DAD molecule as output, thereby ensuring the accuracy of predicting the energy level differences of each molecule to be screened. Then, according to the size of the energy level difference, target molecules that meet the preset ranking are screened from the molecules to be screened to achieve high-throughput screening of molecules. That is, by obtaining the molecular orbital descriptors of each molecule to be screened, the hybridized part in the DAD molecule is simplified, the difficulty of screening molecules according to the hybridized part is reduced, and the preset prediction model is used to improve the accuracy of screening, thereby improving the efficiency of high-throughput screening of molecules.

[0057] It should be noted that the execution subject of this embodiment can be a computing service device with data processing, network communication and program running functions, such as a tablet computer, a personal computer, a mobile phone, etc., or an electronic device capable of realizing the above functions, a molecular screening platform, etc. The following takes the molecular screening platform as an example to illustrate this embodiment and the following embodiments.

[0058] Based on this, the present invention provides a high-throughput screening method for molecules. Figure 1 , Figure 1 This is a schematic diagram of the process of the first embodiment of the high-throughput screening method for molecules of the present application.

[0059] In this embodiment, the high-throughput screening method of molecules includes steps S10 to S30:

[0060] Step S10, obtaining the molecular orbital descriptor of the molecule to be screened;

[0061] It should be noted that the molecules to be screened can be molecules organically combined according to predetermined molecular fragments, and it is necessary to screen out molecules whose physical and chemical properties meet the expectations from these molecules. Molecular orbital descriptors are a mathematical function or symbol system used to characterize the distribution of electron clouds in molecules, and can also predict the properties and reactivity of molecules based on the behavior and interaction of electrons in molecules.

[0062] The molecular orbital descriptor may include, but is not limited to, the HOMO (Highest Occupied Molecular Orbital) descriptor, the LUMO (Lowest Unoccupied Molecular Orbital) descriptor, the HOMOn descriptor, and the LUMOn descriptor of the donor molecule fragment, where HOMOn represents the highest occupied molecular orbital of the nth level, and LUMOn represents the lowest unoccupied molecular orbital of the nth level.

[0063] It can be understood that since molecular orbital descriptors can reflect important information such as the electronic structure of molecules, predicting the energy difference of each molecule to be screened based on the obtained molecular orbital descriptors not only simplifies the difficulty of predicting the energy difference of the molecules to be screened, but also reduces the complexity of screening molecules by using molecular orbital descriptors to replace the hybridized parts in the molecules to predict the energy difference of the molecules, thereby improving the accuracy of obtaining the physicochemical properties of each molecule to be screened.

[0064] In a specific implementation, a specific calculation method can be used to obtain the molecular orbital descriptor of the molecule to be screened. The calculation method can be a calculation software or algorithm based on quantum chemistry principles, for example, a calculation method related to density functional theory (DFT), to accurately calculate the value of the molecular orbital descriptor.

[0065] Step S20, inputting the molecular orbital descriptor into a preset prediction model to predict the energy level difference of each of the molecules to be screened, wherein the preset prediction model is obtained by training a basic machine learning model using the molecular orbital descriptor of each sample DAD molecule as input and the energy level difference of each sample DAD molecule as output;

[0066] It should be noted that the energy level difference is the energy difference between two specific molecular orbitals, which is usually related to the reactivity and stability of the molecule and can be used to evaluate the properties of the molecule. The basic machine learning model can be one or more basic machine learning models such as random forest, XGBoost, K-nearest neighbor regression, decision tree regression, and neural network.

[0067] It can be understood that since the preset prediction model takes the molecular orbital descriptors of each sample DAD molecule as input and the energy level difference of each sample DAD molecule as output, after iterative training of the basic machine learning model, the hyperparameters in the basic machine learning model are adjusted to the optimal level, so that the preset prediction model can accurately extract various physicochemical properties from the molecular orbital descriptors, so as to accurately predict the energy level difference of the corresponding DAD molecule based on the physicochemical properties, thereby improving the accuracy of predicting the energy level difference of the DAD molecule, and since the screening of molecules is based on the size of the energy level difference, therefore, improving the accuracy of the prediction of the energy level difference of the DAD molecule can further improve the accuracy of the molecular screening.

[0068] It can be understood that by combining molecular orbital descriptors with preset prediction models, there is no need to conduct complex experiments to measure the energy level difference of each molecule to be screened, which greatly improves the efficiency of molecular screening and saves time and cost.

[0069] Step S30, based on the order of the energy level differences from large to small, screening target molecules that meet the preset ranking from the molecules to be screened.

[0070] It should be noted that the preset ranking may be an expected value determined based on historical experience, so as to screen out molecules whose performance meets the expectations from the molecules to be screened. The target molecule is a molecule that meets the expectations.

[0071] It can be understood that according to the prediction results output by the preset prediction model, that is, the estimated energy level difference of each molecule to be screened, all candidate objects are arranged in order from high to low, and the top performers within a specific percentage range are selected as the final selected target molecule set to improve the screening rate of molecules and further quickly identify new material candidates with excellent optoelectronic properties.

[0072] In the specific implementation, after obtaining the predicted energy level difference of each molecule to be screened, these molecules are sorted in order from large to small energy level difference. Then, according to the preset ranking requirements, for example, the molecules with the top 10% energy level difference are selected as target molecules. These target molecules show relatively excellent properties in terms of energy level difference and can be used for subsequent research, development or application. For example, in the field of organic optoelectronic materials, molecules with suitable energy level difference may have better photoelectric conversion performance.

[0073] Furthermore, if there are multiple preset prediction models, step S30 further includes:

[0074] The energy level differences predicted by the different preset prediction models are sorted from large to small to obtain corresponding property sorting;

[0075] From the rankings of the properties, extract the molecules to be analyzed that are ranked before a preset ranking;

[0076] Sort the molecules to be analyzed in the property rankings from large to small according to the energy level differences to obtain a comprehensive ranking;

[0077] Based on the comprehensive ranking, the target molecules with the preset ranking are screened from the molecules to be analyzed.

[0078] It should be noted that multiple preset prediction models can be selected. For example, the random forest model and the XGBoost model are used to simultaneously predict the energy level differences of the molecules to be screened. At this time, the property ranking includes the property ranking of the energy level differences predicted by the random forest model and the property ranking of the energy level differences predicted by the XGBoost model. The property ranking is to sort the molecules to be screened according to the energy level differences. There can be multiple rankings for the same molecule in the comprehensive ranking, that is, the ranking ranked according to the energy level differences predicted by the random forest model and the ranking ranked according to the energy level differences predicted by the XGBoost model. If other models are used, it can also include the ranking ranked according to the energy level differences predicted by the other models, etc.

[0079] It is understandable that when there are multiple preset prediction models, the prediction results (energy level differences) of multiple preset prediction models will be obtained. Since the number of molecules to be analyzed is large, while ensuring the accuracy of the energy level difference prediction, the number of energy level differences that need to be further processed in each property ranking can be reduced. That is, the molecules to be analyzed before the preset ranking are first extracted from the property ranking, and then the molecules to be analyzed are comprehensively ranked according to the energy level difference, so as to screen the target molecules with the preset ranking from the comprehensive ranking, thereby reducing the number of molecules to be analyzed and minimizing the number of molecules to be analyzed in the comprehensive ranking, so as to improve the efficiency of screening target molecules.

[0080] It can be understood that since the preset ranking is the final number of target molecules that need to be screened, using the preset ranking to extract the molecules to be analyzed from the ranking of each property can ensure that the molecules to be screened that meet the preset ranking requirements can enter the final screening process, avoiding the situation where some molecules that meet the requirements cannot be selected when screening the molecules to be analyzed based on the comprehensive ranking, thereby improving the accuracy of high-throughput screening molecules.

[0081] It is understandable that since the property rankings of the energy level differences predicted by different models for each molecule to be screened are different, in order to avoid the suddenness or preference of a model prediction, multiple models can be used for prediction, and the property rankings predicted by multiple models can be comprehensively processed, and then the target molecules whose energy differences are ranked before the preset rankings can be obtained from the comprehensive rankings obtained from the comprehensive processing, thereby further improving the accuracy of high-throughput screening of molecules.

[0082] The present embodiment provides a method for high-throughput screening of molecules. Since the molecular orbital descriptors can characterize the behavior of each electron in a molecule, the molecular orbital descriptors of the molecules to be screened are obtained, and then the energy level difference of each molecule to be screened is predicted by using a preset prediction model. With the help of the accuracy of the model prediction, the molecular orbital descriptors are used as the screening basis to predict the energy level difference of each molecule to be screened. The preset prediction model is obtained by training a basic machine learning model with the molecular orbital descriptors of the sample DAD molecules as input and the energy level difference of each sample DAD molecule as output, thereby ensuring the accuracy of predicting the energy level difference of each molecule to be screened. Then, according to the size of the energy level difference, the target molecules that meet the preset ranking are screened from the molecules to be screened to achieve high-throughput screening of molecules. That is, by obtaining the molecular orbital descriptors of each molecule to be screened, the hybrid part in the DAD molecule is simplified, the difficulty of screening molecules according to the hybrid part is reduced, and the preset prediction model is used to improve the accuracy of screening, thereby improving the efficiency of high-throughput screening of molecules.

[0083] Based on the first embodiment of the present application, in the second embodiment of the present application, the same or similar contents as those in the above-mentioned embodiment 1 can refer to the above introduction, and will not be repeated later. Figure 2 Before step S10, the high-throughput screening method for molecules further includes steps S11 to S12:

[0084] Step S11, identifying the donor and the acceptor constituting the molecule to be screened;

[0085] Step S12, based on a preset DFT theory, respectively determining the highest orbital descriptor, the lowest orbital descriptor, the occupied orbital descriptor and the unoccupied orbital descriptor of the donor and the acceptor;

[0086] Among them, the highest orbital descriptor is the descriptor of the highest occupied molecular orbital, the lowest orbital descriptor is the descriptor of the lowest unoccupied molecular orbital, the occupied orbital descriptor is the descriptor of an occupied molecular orbital at a preset energy level, and the unoccupied orbital descriptor is the descriptor of an unoccupied molecular orbital at a preset energy level.

[0087] It should be noted that donors are molecules or atoms that can provide electrons; acceptors are molecules or atoms that can accept electrons. In the study of molecular docking and interaction, identifying donors and acceptors is the key to understanding molecular interactions and the physical and chemical properties of molecules. Occupied orbital descriptors are descriptors of molecular orbitals occupied by electrons at a preset energy level, which can explain the distribution and energy state of electrons in molecules. Unoccupied orbital descriptors are descriptors of molecular orbitals that are not occupied by electrons at a preset energy level, which can explain the distribution and energy state of empty orbitals in molecules.

[0088] It can be understood that identifying the donors and acceptors that constitute the molecules to be screened can obtain the physicochemical properties of the molecules to be screened, for example, the physical structural properties (such as elemental composition, relative molecular mass) of the donors and acceptors, simple chemical properties (such as chemical bonds), and the physical structural properties (such as elemental composition, relative molecular mass), simple chemical properties (such as chemical bonds) of the molecules to be screened, and based on these properties, using the preset DFT theory, the highest orbital descriptor, lowest orbital descriptor, occupied orbital descriptor and unoccupied orbital descriptor of the donor are determined respectively, and the highest orbital descriptor, lowest orbital descriptor, occupied orbital descriptor and unoccupied orbital descriptor of the acceptor are determined, so that the process of obtaining the molecular orbital descriptors of the molecules to be screened is more concise, reducing the chance of error, thereby improving the accuracy of determining the molecular orbital descriptors.

[0089] In the specific implementation, the molecular orbital descriptors of the donor and acceptor fragment structures are calculated using the DFT theoretical calculation method, including eight descriptors of the energy levels of the donor and acceptor, HOMO, LUMO, HOMO1, and LUMO1, and correlation analysis is performed. The linear correlation coefficients between each pair are less than 0.8, which can be directly used in the subsequent machine learning process. Among them, HOMO1 is the occupied orbital descriptor of the first energy level. LUMO1 is the unoccupied orbital descriptor of the first energy level.

[0090] It should be noted that when the target attributes are different, it can be further extended to atomic orbitals such as HOMO2 and LUMO2, so that the prediction results have an ideal overall performance. Among them, the target attribute can be color or conductivity, etc.

[0091] Furthermore, before step S11, the method further includes:

[0092] Retrieving molecular fragments in historical data, and counting the occurrence frequency of each molecular fragment in the historical data;

[0093] Selecting a donor molecule fragment, an acceptor molecule fragment, and a substituent fragment from the molecule fragments based on the occurrence frequency;

[0094] A DAD molecule library storing the molecules to be screened is created based on the donor molecule fragment, the acceptor molecule fragment and the substituent fragment.

[0095] It should be noted that the frequency of occurrence is the number of times a specific molecular fragment is used in historical data.

[0096] It is understandable that by counting the frequency of occurrence of molecular fragments, we can understand which fragments are more common and important in molecular interactions.

[0097] Understandably, a large amount of known DAD-type molecular data was collected from multiple public chemical databases, and these data were preprocessed to extract all unique donor, acceptor and substituent fragments. Then, the frequency of occurrence of these fragments was statistically analyzed, and it was found that certain specific donor and acceptor fragments were more common in high-performance materials. Based on these acceptors and donors, several donor and acceptor fragments with the highest frequency, as well as several common substituent fragments, were selected as the basic units for building new molecules. Finally, these fragments were combined into new DAD-type molecular structures and stored in a DAD molecular library designed specifically for this project, which not only greatly enriched the structural diversity of the molecular library, but also improved the screening efficiency and success rate.

[0098] In the specific implementation, all known molecular fragment information is extracted from the existing historical database. These fragments may include but are not limited to donors, acceptors, and substituents. Then, the frequency of occurrence of these fragments is statistically analyzed using data mining technology to determine which fragments are more common or important in historical topics. Based on the frequency distribution, donor, acceptor, and substituent fragments with higher occurrence frequencies are selected as the basic units for constructing new DAD-type molecules to ensure that the selected molecular fragments are highly representative and practical. Finally, the selected donor, acceptor, and substituent fragments are combined to form a new DAD-type molecular structure, which is then stored in a specially designed molecular library (DAD molecular library). The DAD molecular library not only contains rich structural diversity, but also has high practical application potential.

[0099] For example, refer to Figures 3 to 5 , from the DAD-type molecules used in historical topics, six donors (Acceptor) including PT, MI, DPP, Ant, BZP and BZ; nine acceptors (Donor) including TPA, DPA, PhCz, Pyr, PTZ, TPB, DMF, NPA and Cz; and ten substituents were selected, and a DAD molecular library containing 5,400 molecules to be screened was established using these donors, acceptors and substituents.

[0100] Based on the first and second embodiments of the present application, in the third embodiment of the present application, the same or similar contents as those in the above embodiments can be referred to the above introduction, and will not be described in detail later. Figure 6 Before step S10, the high-throughput screening method of the molecule further includes steps S01 to S04:

[0101] Step S01, selecting a preset number of sample DAD molecules from a DAD molecule library storing the molecules to be screened;

[0102] Step S02, determining the energy level difference of the sample DAD molecules;

[0103] Step S03, based on the energy level difference, selecting a basic machine learning model with the strongest learning ability from a plurality of machine learning models, wherein the basic machine learning models include at least one;

[0104] Step S04, iteratively training the basic machine learning model based on the molecular orbital descriptors of the sample DAD molecules and the corresponding energy level differences to obtain a preset prediction model.

[0105] It should be noted that the sample DAD molecules are a part selected from the molecules to be screened.

[0106] It can be understood that since the donor molecule fragment, the acceptor molecule fragment and the substituent fragment in the sample DAD molecule are parts of the remaining molecules to be screened, by selecting a certain number of molecules from the molecules to be screened as samples to train the machine learning model, the accuracy of the trained model in predicting the energy level difference for the batch of molecules to be screened can be improved.

[0107] It can be understood that by automatically selecting and training the optimal machine learning model from multiple machine learning models as the basic machine learning model for training, the efficiency and accuracy of molecular docking and interaction research are improved.

[0108] It is understandable that since the sample DAD molecules are part of the molecules to be screened, in order to train the model, the commonly used calculation methods will be used to accurately determine the energy level differences of the sample DAD molecules. Therefore, the energy level differences of a part of the molecules to be screened can be determined during the model training process. After the model training is completed, the preset prediction model is used to accurately predict the energy level differences of the remaining molecules to be screened, avoiding the calculation of all the molecules to be screened, thereby improving the efficiency of high-throughput screening of molecules.

[0109] In the specific implementation, 1010 molecules were selected from a database containing 5400 DAD molecules and their energy level differences ΔE TT , or ΔE ST The traditional calculation method was used to calculate the target attribute, and preliminary learning was carried out through basic machine learning models such as random forest, XGBoost, K-nearest neighbor regression, decision tree regression, neural network, etc. The performance results of the preliminary learning were combined to determine that random forest and XGBoost were the strongest, so the hyperparameters were adjusted to calculate the linear correlation coefficient R between the predicted value and the actual value. 2 And root mean square error (RMSE) are used as comprehensive evaluation indicators.

[0110] The main hyperparameters that need to be adjusted in the random forest model include: the number of decision trees (n_estimators), the maximum depth of decision trees (max_depth), the minimum number of separation samples (min_samples_split), the minimum number of leaf node samples (min_samples_leaf), whether to perform random sampling (bootstrap), etc. After random search and grid search, the optimization results are: n_estimators=460, max_depth=130, min_samples_split=3, min_samples_leaf=1, max_features=auto.

[0111] The main hyperparameters that need to be adjusted under the XGBoost model include the number of decision trees (n_estimators), the maximum depth of the decision tree (max_depth), the minimum loss function drop value required for node splitting (gamma), the column sampling ratio (colsample_bytree), the learning rate (learning_rate), the number of samples in the leaf node (min_child_weight), the regularization term parameters (reg_alpha, reg_lambda), the sample ratio used to train the model in each iteration (subsample), etc. After random search and grid search, the optimization results are: n_estimators=900, max_depth=3, gamma=0, colsample_bytree=0.8, learning_rate=0.1, min_child_weight=2, reg_alpha=0.2, reg_lambda=0.05, subsample=0.6.

[0112] Furthermore, step S40 further includes:

[0113] Based on a preset random search method, the basic machine learning model is preliminarily iteratively trained according to the molecular orbital descriptors of the sample DAD molecules and the corresponding energy level differences to obtain a preliminary iterative training model;

[0114] Based on a preset grid search method, the preliminary iterative training model is finally iteratively trained according to the molecular orbital descriptors of the sample DAD molecules and the corresponding energy level differences to obtain a preset prediction model.

[0115] It should be noted that the preset random search method is a method used in the process of adjusting the parameters of machine learning models. It can find the optimal solution by randomly sampling parameter combinations in the parameter space, and can find the approximately optimal parameter combination in a short time. The preset grid search method is a method that systematically traverses all possible parameter combinations to find the strongest parameter settings.

[0116] It is understandable that the time cost of model training is reduced through iterative training based on preset random search and grid search. Since the samples used in the training model are part of the molecules to be screened, the efficiency of molecular screening will be improved by reducing the training time.

[0117] In the specific implementation, five-fold cross validation is used during training to continuously adjust and optimize the parameters of the random forest model and XGBoost model until the comprehensive evaluation index reaches the most ideal state. In the learning process, random search is used for rough adjustment, and then grid search is used for fine adjustment within the range of rough adjustment results, and finally the optimal hyperparameter combination is obtained.

[0118] It should be noted that the above examples are only used to understand the present application and do not constitute a limitation on the high-throughput screening method of the molecules of the present application. More simple transformations based on this technical concept are all within the scope of protection of the present application.

[0119] The present application also provides a high-throughput screening device for molecules, please refer to Figure 7 , the high-throughput screening device for molecules comprises:

[0120] An acquisition module 10 is used to acquire molecular orbital descriptors of the molecules to be screened;

[0121] A prediction module 20, for inputting the molecular orbital descriptor into a preset prediction model to predict the energy level difference of each molecule to be screened, wherein the preset prediction model is obtained by training a basic machine learning model using the molecular orbital descriptor of each sample DAD molecule as input and the energy level difference of each sample DAD molecule as output;

[0122] The screening module 30 is used to screen target molecules that meet a preset ranking from the molecules to be screened based on the order of the energy level differences from large to small.

[0123] Optionally, the acquisition module 10 is also used to identify the donor and the acceptor constituting the molecule to be screened; based on the preset DFT theory, the highest orbital descriptor, the lowest orbital descriptor, the occupied orbital descriptor and the unoccupied orbital descriptor of the donor and the acceptor are determined respectively; wherein the highest orbital descriptor is the descriptor of the highest occupied molecular orbital, the lowest orbital descriptor is the descriptor of the lowest unoccupied molecular orbital, the occupied orbital descriptor is the descriptor of the occupied molecular orbital at a preset energy level, and the unoccupied orbital descriptor is the descriptor of the unoccupied molecular orbital at a preset energy level.

[0124] Optionally, the acquisition module 10 is also used to retrieve molecular fragments in the historical data, and count the occurrence frequency of each molecular fragment in the historical data; select donor molecular fragments, acceptor molecular fragments and substituent fragments from the molecular fragments based on the occurrence frequency; and create a DAD molecular library for storing the molecules to be screened based on the donor molecular fragments, the acceptor molecular fragments and the substituent fragments.

[0125] Optionally, the prediction module 20 is also used to select a preset number of sample DAD molecules from the DAD molecule library storing the molecules to be screened; determine the energy level difference of the sample DAD molecules; based on the energy level difference, select a basic machine learning model with the strongest learning ability from multiple machine learning models, and the basic machine learning model includes at least one; based on the molecular orbital descriptors of the sample DAD molecules and the corresponding energy level differences, iteratively train the basic machine learning model to obtain a preset prediction model.

[0126] Optionally, the prediction module 20 is also used to perform preliminary iterative training on the basic machine learning model based on a preset random search method, according to the molecular orbital descriptors of the sample DAD molecules and the corresponding energy level differences, to obtain a preliminary iterative training model; based on a preset grid search method, perform final iterative training on the preliminary iterative training model according to the molecular orbital descriptors of the sample DAD molecules and the corresponding energy level differences, to obtain a preset prediction model.

[0127] Optionally, if there are multiple preset prediction models, the screening module 30 is also used to sort the energy level differences predicted by different preset prediction models from large to small to obtain corresponding property rankings; extract the molecules to be analyzed before the preset ranking from each of the property rankings; sort the molecules to be analyzed in each of the property rankings from large to small according to the energy level differences to obtain a comprehensive ranking; based on the comprehensive ranking, screen the target molecules of the preset ranking from each of the molecules to be analyzed.

[0128] The high-throughput screening device for molecules provided in the present application adopts the high-throughput screening method for molecules in the above-mentioned embodiments, and can solve the technical problem of low efficiency of high-throughput screening for molecules. Compared with the prior art, the beneficial effects of the high-throughput screening device for molecules provided in the present application are the same as the beneficial effects of the high-throughput screening method for molecules provided in the above-mentioned embodiments, and other technical features in the high-throughput screening device for molecules are the same as the features disclosed in the above-mentioned embodiment methods, which will not be described in detail here.

[0129] The present application provides a high-throughput screening device for molecules, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the high-throughput screening method for molecules in the above-mentioned embodiment one.

[0130] Reference below Figure 8 , which shows a schematic diagram of the structure of a high-throughput screening device for molecules suitable for implementing the embodiments of the present application. The high-throughput screening device for molecules in the embodiments of the present application may include, but is not limited to, mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 8 The high-throughput screening device for molecules shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0131] like Figure 8As shown, the high-throughput screening device for molecules may include a processing device 1001 (such as a central processing unit, a graphics processor, etc.), which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM: Read Only Memory) 1002 or the program loaded from the storage device 1003 to the random access memory (RAM: Random Access Memory) 1004. In RAM1004, various programs and data required for the operation of the high-throughput screening device for molecules are also stored. The processing device 1001, ROM1002 and RAM1004 are connected to each other through a bus 1005. The input / output (I / O) interface 1006 is also connected to the bus. Generally, the following systems can be connected to the I / O interface 1006: an input device 1007 including, for example, a touch screen, a touch pad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD: Liquid Crystal Display), a speaker, a vibrator, etc.; a storage device 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1009. Communication device 1009 can allow the high-throughput screening device of molecules to communicate wirelessly or wired with other devices to exchange data. Although the figure shows a high-throughput screening device of molecules with various systems, it should be understood that it is not required to implement or have all the systems shown. More or fewer systems can be implemented or possessed alternatively.

[0132] In particular, according to the embodiments disclosed in the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present application include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device, or installed from a storage device 1003, or installed from a ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present application are executed.

[0133] The high-throughput screening device for molecules provided in the present application adopts the high-throughput screening method for molecules in the above-mentioned embodiment, which can solve the technical problem of low efficiency of high-throughput screening for molecules. Compared with the prior art, the beneficial effects of the high-throughput screening device for molecules provided in the present application are the same as the beneficial effects of the high-throughput screening method for molecules provided in the above-mentioned embodiment, and other technical features in the high-throughput screening device for molecules are the same as the features disclosed in the method of the previous embodiment, which will not be described in detail here.

[0134] It should be understood that the various parts disclosed in this application can be implemented by hardware, software, firmware or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0135] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0136] The present application provides a computer-readable storage medium having computer-readable program instructions (ie, computer program) stored thereon, wherein the computer-readable program instructions are used to execute the high-throughput screening method for molecules in the above-mentioned embodiments.

[0137] The computer-readable storage medium provided in the present application may be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, systems or devices, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM: Random Access Memory), a read-only memory (ROM: Read Only Memory), an erasable programmable read-only memory (EPROM: Erasable Programmable Read Only Memory or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM: CD-Read Only Memory), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, system or device. The program code contained on the computer-readable storage medium may be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (Radio Frequency: Radio Frequency), etc., or any suitable combination of the above.

[0138] The computer-readable storage medium may be included in the high-throughput screening device for molecules; or may exist independently without being assembled into the high-throughput screening device for molecules.

[0139] The above-mentioned computer-readable storage medium carries one or more programs. When the above-mentioned one or more programs are executed by the high-throughput screening device for molecules, the high-throughput screening device for molecules enables the following: to obtain the molecular orbital descriptors of the molecules to be screened; to input the molecular orbital descriptors into a preset prediction model to predict the energy level differences of each of the molecules to be screened, wherein the preset prediction model is obtained by training a basic machine learning model with the molecular orbital descriptors of each sample DAD molecule as input and the energy level differences of each sample DAD molecule as output; and to screen the target molecules that meet the preset ranking from each of the molecules to be screened based on the ordering of the energy level differences from large to small.

[0140] Computer program code for performing the operations of the present application may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and 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 through 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., via the Internet using an Internet service provider).

[0141] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present application. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession 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 square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0142] The modules involved in the embodiments described in this application may be implemented by software or hardware, wherein the name of the module does not constitute a limitation on the unit itself in some cases.

[0143] The readable storage medium provided in the present application is a computer-readable storage medium, which stores computer-readable program instructions (i.e., computer programs) for executing the high-throughput screening method for the above-mentioned molecules, and can solve the technical problem of low efficiency in high-throughput screening of molecules. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in the present application are the same as the beneficial effects of the high-throughput screening method for molecules provided in the above-mentioned embodiments, and will not be described in detail here.

[0144] The present application also provides a computer program product, comprising a computer program, which implements the steps of the high-throughput screening method for molecules as described above when executed by a processor.

[0145] The computer program product provided in this application can solve the technical problem of low efficiency in high-throughput screening of molecules. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as the beneficial effects of the high-throughput screening method of molecules provided in the above embodiments, which will not be repeated here.

[0146] The above descriptions are only some embodiments of the present application, and are not intended to limit the patent scope of the present application. All equivalent structural changes made using the contents of the present application specification and drawings under the technical concept of the present application, or direct / indirect applications in other related technical fields are included in the patent protection scope of the present application.

Claims

1. A high-throughput screening method for molecules, characterized in that: The method includes: Obtaining a molecular orbital descriptor of the molecule to be screened, wherein the molecular orbital descriptor is a mathematical function or symbol system used to characterize the distribution of electron clouds in a molecule and is also used to predict the properties and reactivity of the molecule based on the behavior and interaction of electrons in the molecule; Inputting the molecular orbital descriptor into a preset prediction model to predict the energy level difference of each molecule to be screened, wherein the preset prediction model is obtained by training a basic machine learning model using the molecular orbital descriptor of each sample DAD molecule as input and the energy level difference of each sample DAD molecule as output; Based on the order of the energy level differences from large to small, target molecules that meet the preset ranking are screened from the molecules to be screened.

2. The method according to claim 1, characterized in that The step of obtaining the molecular orbital descriptor of the molecule to be screened comprises: Identify the donors and acceptors that make up the molecule to be screened; Based on a preset DFT theory, determining the highest orbital descriptor, the lowest orbital descriptor, the occupied orbital descriptor and the unoccupied orbital descriptor of the donor and the acceptor respectively; Among them, the highest orbital descriptor is the descriptor of the highest occupied molecular orbital, the lowest orbital descriptor is the descriptor of the lowest unoccupied molecular orbital, the occupied orbital descriptor is the descriptor of an occupied molecular orbital at a preset energy level, and the unoccupied orbital descriptor is the descriptor of an unoccupied molecular orbital at a preset energy level.

3. The method according to claim 2, characterized in that The step of identifying the donor and the acceptor constituting the molecule to be screened also includes: Retrieving molecular fragments in historical data, and counting the occurrence frequency of each molecular fragment in the historical data; Selecting a donor molecule fragment, an acceptor molecule fragment, and a substituent fragment from the molecule fragments based on the occurrence frequency; A DAD molecule library storing the molecules to be screened is created based on the donor molecule fragment, the acceptor molecule fragment and the substituent fragment.

4. The method according to any one of claims 1 to 3, characterized in that The step of obtaining the molecular orbital descriptor of the molecule to be screened also includes: Selecting a preset number of sample DAD molecules from a DAD molecule library storing the molecules to be screened; determining the energy level difference of the sample DAD molecules; Based on the energy level difference, selecting a basic machine learning model with the strongest learning ability from a plurality of machine learning models, wherein the basic machine learning models include at least one; Based on the molecular orbital descriptors of the sample DAD molecules and the corresponding energy level differences, the basic machine learning model is iteratively trained to obtain a preset prediction model.

5. The method according to claim 4, characterized in that The step of iteratively training the basic machine learning model based on the molecular orbital descriptors of the sample DAD molecules and the corresponding energy level differences to obtain a preset prediction model comprises: Based on a preset random search method, the basic machine learning model is preliminarily iteratively trained according to the molecular orbital descriptors of the sample DAD molecules and the corresponding energy level differences to obtain a preliminary iterative training model; Based on a preset grid search method, the preliminary iterative training model is finally iteratively trained according to the molecular orbital descriptors of the sample DAD molecules and the corresponding energy level differences to obtain a preset prediction model.

6. The method according to claim 1, characterized in that If there are multiple preset prediction models, the step of selecting target molecules that meet the preset ranking from the molecules to be screened based on the order of the energy level differences from large to small includes: The energy level differences predicted by the different preset prediction models are sorted from large to small to obtain corresponding property sorting; From the rankings of the properties, extract the molecules to be analyzed that are ranked before a preset ranking; Sort the molecules to be analyzed in the property rankings from large to small according to the energy level differences to obtain a comprehensive ranking; Based on the comprehensive ranking, the target molecules with the preset ranking are screened from the molecules to be analyzed.

7. A high-throughput screening device for molecules, characterized in that: The device comprises: An acquisition module, used to acquire the molecular orbital descriptor of the molecule to be screened, wherein the molecular orbital descriptor is a mathematical function or symbol system used to characterize the distribution of the electron cloud in the molecule, and is also used to predict the properties and reactivity of the molecule based on the behavior and interaction of the electrons in the molecule; A prediction module is used to input the molecular orbital descriptor into a preset prediction model to predict the energy level difference of each molecule to be screened, wherein the preset prediction model is obtained by training a basic machine learning model with the molecular orbital descriptor of each sample DAD molecule as input and the energy level difference of each sample DAD molecule as output; a screening module is used to screen target molecules that meet the preset ranking from each molecule to be screened based on the order of the energy level difference from large to small.

8. A high-throughput screening device for molecules, characterized in that: The device comprises: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the high-throughput screening method for molecules according to any one of claims 1 to 6.

9. A storage medium, characterized in that: The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the high-throughput screening method for molecules according to any one of claims 1 to 6 are implemented.

10. A computer program product, characterized in that The computer program product comprises a computer program, which, when executed by a processor, implements the steps of the high-throughput screening method for molecules according to any one of claims 1 to 6.

Images