Composite structure prediction method, model training method and related equipment

Abstract

The invention provides a structure prediction method of a compound, a model training method and related equipment, and relates to the technical field of artificial intelligence, in particular to the technical fields of deep learning, biological calculation and the like. According to the scheme, a biomolecule sequence combination is obtained; wherein the biomolecule sequence combination is obtained by performing sequence combination according to a plurality of specified biomolecule sequences; predicting a first probability distribution of the biomolecular sequence combination; the first probability distribution is used for indicating first probabilities of a plurality of candidate structural unit groups in the biomolecule sequence combination, each candidate structural unit group comprises structural units in at least two biomolecule sequences, and the first probabilities are used for indicating the possibility of interaction between the structural units in the candidate structural unit groups; determining at least one first structural unit group from the plurality of candidate structural unit groups according to the first probability distribution; and predicting a target structure of the biomolecular compound according to the structural units interacting with each other in each first structural unit group.

Description

Technical Field

[0001] The present disclosure relates to the field of artificial intelligence technology, specifically to technical fields such as deep learning and biocomputing, and in particular to a structure prediction method, a model training method, and related equipment for a complex. Background Art

[0002] With the rapid development of computer technology and bioinformatics, biomolecular complexes play an irreplaceable role as the basic units for performing key biological functions. In biology, multiple biomolecular sequences (such as proteins, nucleic acids, etc.) form complexes through precise and dynamic interactions to achieve core life activities such as signal transduction, gene regulation, and metabolic catalysis. However, due to the high diversity of biomolecular sequences and the complexity of their interactions, accurately predicting and designing the three-dimensional structure of biomolecular complexes remains one of the most challenging problems in the current fields of computational biology and structural biology. Summary of the Invention

[0003] The present disclosure provides a structure prediction method, a model training method and related equipment for a complex.

[0004] According to one aspect of the present disclosure, a method for predicting the structure of a complex is provided, comprising:

[0005] Obtaining a biomolecule sequence combination; wherein the biomolecule sequence combination is obtained by sequence combination based on a plurality of specified biomolecule sequences;

[0006] Predicting a first probability distribution of the biomolecule sequence combination; wherein the first probability distribution is used to indicate first probabilities of a plurality of candidate structural unit groups in the biomolecule sequence combination, the candidate structural unit groups comprising structural units from at least two of the biomolecule sequences, and the first probabilities are used to indicate likelihoods of interactions between structural units in the candidate structural unit groups;

[0007] determining at least one first structural unit group from the plurality of candidate structural unit groups according to the first probability distribution;

[0008] A target structure of the biomolecule complex is predicted based on the interacting structural units in each of the first structural unit groups.

[0009] According to another aspect of the present disclosure, a method for training an all-atom structure prediction model is provided, comprising:

[0010] Acquire a training sample; wherein the training sample includes a sample biomolecule sequence combination, and the sample biomolecule sequence combination is obtained based on a plurality of sample biomolecule sequence combinations;

[0011] A structure prediction model is used to predict a second probability distribution of the sample biomolecule sequence combination, and a predicted structure of the biomolecule complex is generated based on the second probability distribution; wherein the second probability distribution is used to indicate second probabilities of a plurality of candidate structural unit groups in the sample biomolecule sequence combination, the candidate structural unit groups including structural units in at least two of the sample biomolecule sequences, and the second probabilities are used to indicate the likelihood of interaction between the structural units in the candidate structural unit groups;

[0012] The structure prediction model is trained based on the differences between the annotated structures corresponding to the plurality of sample biomolecule sequences and the predicted structures.

[0013] According to another aspect of the present disclosure, there is provided a structure prediction apparatus for a complex, comprising:

[0014] An acquisition module, configured to acquire a biomolecule sequence combination; wherein the biomolecule sequence combination is obtained by combining a plurality of specified biomolecule sequences;

[0015] a prediction module configured to predict a first probability distribution of the biomolecule sequence combination; wherein the probability distribution indicates first probabilities of a plurality of candidate structural unit groups in the biomolecule sequence combination, the candidate structural unit groups comprising structural units from at least two of the biomolecule sequences, and the probabilities indicating likelihood of interaction between structural units in the candidate structural unit groups;

[0016] a determining module, configured to determine at least one first structural unit group from the plurality of candidate structural unit groups according to the first probability distribution;

[0017] A generating module is used to predict a target structure of a biomolecule complex based on the interacting structural units in each of the first structural unit groups.

[0018] According to another aspect of the present disclosure, a training device for a structure prediction model is provided, comprising:

[0019] An acquisition module, configured to acquire a training sample; wherein the training sample comprises a sample biomolecule sequence combination, and the sample biomolecule sequence combination is obtained based on a plurality of sample biomolecule sequence combinations;

[0020] a prediction module configured to predict a second probability distribution of the sample biomolecule sequence combination using an all-atom structure prediction model, and generate a predicted structure of the biomolecule complex based on the second probability distribution; wherein the second probability distribution is configured to indicate second probabilities of a plurality of candidate structural unit groups in the sample biomolecule sequence combination, the candidate structural unit groups comprising structural units from at least two of the sample biomolecule sequences, and the second probabilities are configured to indicate likelihoods of interaction between structural units in the candidate structural unit groups;

[0021] A training module is used to train the structure prediction model according to the differences between the annotated structures corresponding to the multiple sample biomolecule sequences and the predicted structures.

[0022] According to another aspect of the present disclosure, there is provided an electronic device, including:

[0023] at least one processor; and

[0024] a memory communicatively connected to the at least one processor; wherein,

[0025] The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the method proposed in the above-mentioned first aspect of the present disclosure, or to execute the method proposed in the above-mentioned other aspect of the present disclosure.

[0026] According to another aspect of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to enable the computer to execute the method proposed in the above-mentioned first aspect of the present disclosure, or to execute the method proposed in the above-mentioned other aspect of the present disclosure.

[0027] According to another aspect of the present disclosure, a computer program product is provided, including a computer program, which, when executed by a processor, implements the method proposed in the above-mentioned first aspect of the present disclosure, or executes the method proposed in the above-mentioned other aspect of the present disclosure.

[0028] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present disclosure, nor are they intended to limit the scope of the present disclosure. Other features of the present disclosure will become readily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] The accompanying drawings are provided to facilitate a better understanding of the present invention and do not constitute a limitation of the present disclosure.

[0030] Figure 1 A schematic flow chart of the structure prediction method for a complex provided in Example 1 of the present disclosure;

[0031] Figure 2 A schematic flow chart of the structure prediction method for a complex provided in Example 2 of the present disclosure;

[0032] Figure 3 A schematic diagram of the flow chart of the structure prediction method for the complex provided in Example 3 of the present disclosure;

[0033] Figure 4 A schematic flow chart of the structure prediction method for the complex provided in Example 4 of the present disclosure;

[0034] Figure 5 A flowchart of a method for training a structure prediction model provided in the fifth embodiment of the present disclosure;

[0035] Figure 6 A schematic diagram illustrating the principles of a training method for a structure prediction model provided in an embodiment of the present disclosure;

[0036] Figure 7 This is a schematic diagram of the structure of the complex structure prediction device provided in Example 6 of the present disclosure;

[0037] Figure 8 A schematic diagram of the structure of a training device for a structure prediction model provided in the seventh embodiment of the present disclosure;

[0038] Figure 9 A schematic block diagram of an example electronic device that can be used to implement embodiments of the present disclosure is shown. DETAILED DESCRIPTION

[0039] The following description of exemplary embodiments of the present disclosure is made in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding. These details should be considered as merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0040] In recent years, with the rapid development of artificial intelligence (AI) technology, AI-based structure prediction methods for biomolecular complexes (including proteins, small molecules, RNA, DNA, and other types) have attracted widespread attention from academia and industry compared to traditional experimental techniques. How to use deep learning technology to improve the accuracy and efficiency of structural prediction of biomolecular complexes has become a core issue in current research. Its application scenarios are wide-ranging, including but not limited to protein three-dimensional structure prediction, protein-small molecule interaction conformation prediction, RNA structure prediction, etc. Specifically, given a biomolecule sequence (such as protein sequence, RNA sequence, DNA sequence), small molecule molecular formula, covalent information, protein modification information, etc., it is necessary to predict the atomic-level 3D spatial structure of these biomolecules, including their own structure and binding posture. If an accurate structure can be obtained, it will play an important role in biologists' analysis of biological processes and downstream applications such as affinity prediction, drug screening, and drug design.

[0041] In related technologies, all-atom structure prediction models are used to predict the structure of biomolecular complexes. Among them, all-atom structure prediction models (such as AlphaFold3) usually contain two networks: a structure generation network and a structure scoring network. The structure generation network is used to generate the corresponding conformation based on the input biomolecular sequence information; the structure scoring network is used to score the generated conformation to evaluate the quality of the generated conformation; in the reasoning process of generating the complex, the structure generation network will be called multiple times to generate different conformations, and then these conformations will be scored using the structure scoring network, and the conformation with the highest score will be selected as the final predicted conformation. In order to make the different conformations generated different, the following methods are usually relied on:

[0042] (1) Perform data perturbation on the input, such as perturbing the multiple sequence alignment (MSA) information in the input;

[0043] (2) Perturb the network structure, such as turning on the network’s random dropout module;

[0044] (3) The structure generation module is based on a diffusion framework, and its output itself is random. Different diffusion random seeds may lead to different conformations.

[0045] Although related sampling technologies can provide different conformations, their sampling is highly random. For example, highly similar conformations may be repeatedly sampled, or it may be difficult to sample the correct conformation, resulting in low efficiency.

[0046] In response to at least one of the above-mentioned problems, the present disclosure provides a structure prediction method, a model training method, and related equipment for a complex.

[0047] The structure prediction method, model training method, and related devices of the composite according to the present disclosure are described below with reference to the accompanying drawings.

[0048] Figure 1 Schematic diagram of the process of the structure prediction method of the complex provided in Example 1 of the present disclosure.

[0049] The embodiment of the present disclosure takes the structure prediction method of the complex as an example, in which the structure prediction device of the complex is configured. The structure prediction device of the complex can be applied to any electronic device so that the electronic device can perform the structure prediction function of the complex.

[0050] Among them, the electronic device can be any device with computing capabilities, such as a computer, mobile terminal, server, etc. The mobile terminal can be, for example, a vehicle-mounted device, a mobile phone, a tablet computer, a personal digital assistant, a wearable device, and other hardware devices with various operating systems, touch screens and / or display screens.

[0051] like Figure 1 As shown, the structure prediction method of the complex may include the following steps:

[0052] Step 101: Obtain a biomolecule sequence combination.

[0053] The biomolecule sequence combination is obtained by combining a plurality of specified biomolecule sequences.

[0054] To effectively simulate potential interaction patterns between different biomolecule sequences, in the disclosed embodiments, multiple designated biomolecule sequences are paired or combined in various ways to further analyze potential interactions between them. For example, the multiple designated biomolecule sequences may include a protein sequence and a nucleic acid sequence, or two or more different protein sequences.

[0055] Step 102: predict a first probability distribution of biomolecule sequence combinations.

[0056] The first probability distribution is used to indicate a first probability of a plurality of candidate structural unit groups in the biomolecule sequence combination, wherein the candidate structural unit groups include structural units in at least two biomolecule sequences, and the first probability is used to indicate a possibility of interaction between the structural units in the candidate structural unit groups.

[0057] It should be understood that each biomolecule sequence may include multiple structural units (also called tokens). For example, for protein sequences, DNA sequences, and RNA sequences, each residue is considered a token. In the embodiments of the present disclosure, for each biomolecule sequence combination, a probability distribution (also called a first probability distribution) is further predicted to evaluate which structural units in the biomolecule sequence combination are most likely to interact with each other.

[0058] Step 103 : Determine at least one first structural unit group from a plurality of candidate structural unit groups according to the first probability distribution.

[0059] In order to accurately determine the structural unit group that is most likely to interact from multiple candidate structural unit groups in the biomolecule sequence combination, as a possible implementation method, at least one first structural unit group is determined from the multiple candidate structural unit groups based on a first probability of the multiple candidate structural unit groups in the biomolecule sequence combination indicated by a first probability distribution.

[0060] Step 104 : predicting a target structure of the biomolecule complex based on the interacting structural units in each first structural unit group.

[0061] Furthermore, based on the relevant information of the first structural unit group, the target structure of the biomolecular complex is predicted; wherein the relevant information of the first structural unit group includes but is not limited to detailed information such as the position, type, interaction mode, etc. of the structural units.

[0062] For example, if the first building block group contains specific amino acid residues (such as lysine and aspartic acid) and it is known that there may be non-covalent interactions between them (such as salt bridges), such interactions will be specifically considered when predicting the complex structure.

[0063] In summary, by obtaining a biomolecular sequence combination composed of multiple specified biomolecular sequences, the potential interaction patterns between different biomolecules are simulated, providing a rich candidate space for subsequent structure prediction; then, based on the first probability distribution of the predicted biomolecular sequence combination, the possibility of interaction between the structural units in each candidate structural unit group in the biomolecular sequence combination is quantitatively evaluated, thereby screening out the first structural unit group with the greatest possibility of interaction, avoiding blind search of massive conformational space; finally, based on the information of the interacting structural units in the first structural unit group, the overall three-dimensional structure of the target biomolecular complex is further predicted, and the interacting structural units in the first structural unit group are used as a guide to ensure that the structure of the generated biomolecular complex meets its interaction requirements, thereby improving the reliability and biological rationality of the structure prediction.

[0064] In order to clearly illustrate how any embodiment of the present disclosure predicts the target structure of a biomolecular complex based on the interacting structural units in each first structural unit group, the present disclosure proposes another complex structure prediction method.

[0065] Figure 2 This is a schematic flow chart of the structure prediction method for the complex provided in Example 2 of the present disclosure.

[0066] like Figure 2 As shown, the structure prediction method of the complex may include the following steps:

[0067] Step 201: Obtain a biomolecule sequence combination.

[0068] The biomolecule sequence combination is obtained by combining a plurality of specified biomolecule sequences.

[0069] Step 202: predict a first probability distribution of biomolecule sequence combinations.

[0070] The first probability distribution is used to indicate a first probability of a plurality of candidate structural unit groups in the biomolecule sequence combination, wherein the candidate structural unit groups include structural units in at least two biomolecule sequences, and the first probability is used to indicate a possibility of interaction between the structural units in the candidate structural unit groups.

[0071] Step 203: Determine at least one first structural unit group from a plurality of candidate structural unit groups according to the first probability distribution.

[0072] In order to accurately screen out the structural unit groups that are most likely to interact, as a possible implementation method, based on the first probability of each candidate structural unit group in the first probability distribution, the first structural unit group that is most likely to interact is determined from each candidate structural unit group.

[0073] As an example, at least one first structural unit group is determined from each candidate structural unit group according to the first probability of each candidate structural unit group in the first probability distribution; wherein the first probability of the first structural unit group is greater than a set threshold.

[0074] That is to say, a threshold is set as the screening criterion, and only candidate structural unit groups whose first probability exceeds this threshold will be selected as the first structural unit group, ensuring that the selected first structural unit group has high credibility and stability, effectively excluding low-probability or unstable structural unit groups, thereby improving the accuracy and reliability of subsequent biomolecular complex structure predictions.

[0075] Step 204: Obtain the similarity between any two first structural unit groups.

[0076] The similarity is used to indicate the degree of similarity between biomolecule complexes generated based on any two interacting structural units in the first structural unit group.

[0077] In order to avoid the possible repeated sampling of highly similar conformations, as a possible implementation method, the similarity between any two first structural unit groups is determined, and the first structural unit groups are screened based on the similarity between any two first structural unit groups. Therefore, the similarity between any two first structural unit groups can be obtained first.

[0078] As an example, structure prediction is performed based on any two first structural unit groups to obtain predicted structures of any two first structural unit groups; wherein the predicted structures are used to indicate the structures of the biomolecule complexes generated based on the first structural unit groups; based on the predicted structures of any two first structural unit groups, the degree of similarity between the structures of the biomolecule complexes generated by any two first structural unit groups is evaluated; and based on the degree of similarity, the degree of similarity between any two first structural unit groups is determined.

[0079] That is, for each pair of selected first structural unit groups (i.e., a combination of at least two structural units), computational biology tools or algorithms are used to predict the overall three-dimensional structure (predicted structure) of the biomolecular complex that they may form, and then the spatial overlap, atomic position differences, and other geometric and topological features of the two predicted structures are compared to determine the similarity scores of the two predicted structures, wherein the similarity score is used to indicate the degree of similarity between the structures of the biomolecular complexes generated by any two first structural unit groups, and based on the similarity score, the similarity between any two first structural unit groups is determined, wherein it should be noted that the higher the degree of similarity between the structures of the biomolecular complexes generated by any two first structural unit groups, the higher the similarity between any two first structural unit groups.

[0080] Step 205 : Filter the first structural unit groups based on the similarities between the first structural unit groups to obtain retained first structural unit groups.

[0081] In order to remove highly similar structural unit groups, reduce unnecessary repeated calculations, and save time and computing resources, as a possible implementation method, the first structural unit groups are filtered according to a set similarity threshold. If it is found that some first structural unit groups are very similar (i.e., exceeding the set threshold), one representative first structural unit group can be retained, while other similar first structural unit groups are removed, thereby obtaining the retained first structural unit groups.

[0082] Step 206 : predicting a target structure of the biomolecule complex based on the interacting structural units in the retained first structural unit group.

[0083] It should be understood that since the retained first structural unit group represents different interaction patterns and structural features, in order to improve the diversity of the predicted structures of the biomolecular complex, the target structure of the biomolecular complex is predicted based on the relevant information of the interacting structural units in the retained first structural unit group.

[0084] It should be noted that the execution process of steps 201 to 202 can be implemented in any of the embodiments of the present disclosure, and the embodiments of the present disclosure do not limit this and will not be described in detail.

[0085] In summary, by calculating the similarity between any two first structure unit groups and filtering all first structure unit groups based on the similarity, highly similar or repeated first structure unit groups can be effectively removed, thereby retaining more representative and diverse first structure unit groups. This process not only reduces redundant calculations in subsequent structure predictions and improves computational efficiency, but also avoids falling into local optimality or repeated sampling of similar conformations. Furthermore, predicting the target structure of biomolecular complexes based on the interacting structural units in the retained first structure unit groups helps to focus on key interaction patterns and improve the accuracy and biological rationality of the prediction results.

[0086] In order to clearly illustrate how any embodiment of the present disclosure predicts the target structure of a biomolecular complex based on the interacting structural units in the retained first structural unit group, the present disclosure proposes another structural prediction method for the complex.

[0087] Figure 3 This is a schematic flow chart of the structure prediction method for the complex provided in Example 3 of the present disclosure.

[0088] like Figure 3 As shown, the structure prediction method of the complex may include the following steps:

[0089] Step 301: Obtain a biomolecule sequence combination.

[0090] The biomolecule sequence combination is obtained by combining a plurality of specified biomolecule sequences.

[0091] Step 302: predict a first probability distribution of biomolecule sequence combinations.

[0092] The first probability distribution is used to indicate a first probability of a plurality of candidate structural unit groups in the biomolecule sequence combination, wherein the candidate structural unit groups include structural units in at least two biomolecule sequences, and the first probability is used to indicate a possibility of interaction between the structural units in the candidate structural unit groups.

[0093] Step 303: Determine at least one first structural unit group from a plurality of candidate structural unit groups according to the first probability distribution.

[0094] Step 304: Obtain the similarity between any two first structural unit groups.

[0095] The similarity is used to indicate the degree of similarity between biomolecule complexes generated based on any two interacting structural units in the first structural unit group.

[0096] Step 305 : Filter the first structural unit groups based on the similarities between the first structural unit groups to obtain retained first structural unit groups.

[0097] Step 306 : predicting candidate structures of biomolecular complexes corresponding to each retained first structural unit group based on the interacting structural units in each retained first structural unit group.

[0098] In order to improve computational efficiency and result diversity and avoid repeated sampling, as a possible implementation method, multiple rounds of sampling are performed on the retained first structural unit group, and in each round, an unlabeled structural unit group is selected from it for labeling, and the structure of the corresponding biomolecular complex is generated based on the interacting structural units in the sampled structural unit group.

[0099] As an example, multiple rounds of sampling process are performed on the retained first structural unit group; wherein the sampling process includes: sampling an unlabeled first structural unit group from the retained first structural unit group, and labeling a first structural unit group; wherein the label is used to indicate that a first structural unit group has been sampled; and generating a candidate structure based on the interacting structural units in a first structural unit group.

[0100] That is to say, in each round of sampling, a structural unit group is randomly or according to a certain strategy from the first structural unit group that has not been labeled (i.e., has not been processed). Once a first structural unit group is selected, it will be marked. The purpose of the mark is to indicate that the structural unit group has been sampled and processed, so as to avoid repeated processing of the same structural unit group in the same round or multiple rounds of sampling. For each sampled and labeled first structural unit group, a structure of a biomolecular complex, i.e., a candidate structure, will be generated based on the interaction information between its structural units.

[0101] Step 307: predict the structure score of each candidate structure.

[0102] The structure score is used to indicate the degree of matching between the candidate structure and the corresponding structural unit of the first structural unit group.

[0103] To effectively evaluate the plausibility of each candidate structure and screen for the optimal conformation, a possible approach is to evaluate the degree of match between the candidate structure of the biomolecular complex and the structural units in its corresponding first building block group, thereby obtaining a structural score for each candidate structure. For example, structural scores can be calculated using various methods, such as energy functions, geometric matching, and comparison with known similar structures. It should be noted that a higher structural score indicates that the candidate structure is more consistent with the expected structural unit combination and its interaction pattern.

[0104] Step 308: Determine the target structure from the candidate structures based on the structure scores of the candidate structures.

[0105] In order to improve the quality of biomolecular complex structure prediction, as a possible implementation method, all candidate structures are ranked according to their scores, and the candidate structure with the highest score can be directly selected as the only target structure.

[0106] It should be noted that the execution process of steps 301 to 305 can be implemented in any of the embodiments of the present disclosure, and the embodiments of the present disclosure do not limit this and will not be described in detail.

[0107] In summary, by predicting candidate structures of biomolecular complexes based on the interacting structural units in the retained first structural unit group, we can fully explore potential complex conformations, ensure that all possible structural patterns are evaluated, and avoid missing any potential important structural information; then, a structural score is calculated for each generated candidate structure, which reflects the degree of match and stability between the candidate structure and its corresponding first structural unit group; finally, according to the calculated structural score, the target structure is determined from each candidate structure, thereby improving the quality and accuracy of biomolecular complex structure prediction.

[0108] In order to clearly illustrate how to predict the probability distribution of biomolecular sequence combinations in any embodiment of the present disclosure, the present disclosure proposes another complex structure prediction method.

[0109] Figure 4 This is a schematic flow chart of the structure prediction method for the complex provided in Example 4 of the present disclosure.

[0110] like Figure 4 As shown, the structure prediction method of the complex may include the following steps:

[0111] Step 401: Obtain a biomolecule sequence combination.

[0112] The biomolecule sequence combination is obtained by combining a plurality of specified biomolecule sequences.

[0113] Step 402: Acquire multiple candidate structural unit groups in the biomolecule sequence combination.

[0114] In the disclosed embodiments, a candidate structural unit group refers to a collection of small molecular units that are identified as potentially forming stable complexes within a biomolecule sequence based on certain criteria (e.g., physicochemical properties, known interaction databases, etc.). Each candidate structural unit group contains at least two structural units (e.g., amino acid residues or nucleotides), and these units may interact with each other.

[0115] Step 403: Calculate the spatial distances between multiple structural units in any candidate structural unit group.

[0116] In order to accurately determine whether multiple structural units in any candidate structural unit group are physically "close" and thus have the potential to interact with each other, as a possible implementation method, the spatial distance between multiple structural units in any candidate structural unit group is calculated using methods such as the Euclidean distance formula.

[0117] Step 404 : determining a first probability of any candidate structural unit group according to the spatial distances between multiple structural units in any candidate structural unit group.

[0118] Among them, the first probability is negatively correlated with the spatial distance.

[0119] As a possible implementation method, the greater the spatial distance between multiple structural units in any candidate structural unit group, the smaller the first probability of the candidate structural unit group, that is, the smaller the possibility of interaction between the structural units in the candidate structural unit group; the smaller the spatial distance between multiple structural units in any candidate structural unit group, the greater the first probability of the candidate structural unit group, that is, the greater the possibility of interaction between the structural units in the candidate structural unit group.

[0120] Step 405 : generating a first probability distribution of biomolecule sequence combinations based on the first probabilities of the candidate structural unit groups.

[0121] Furthermore, the first probabilities of all candidate structural unit groups are integrated into a comprehensive probability distribution map, namely the first probability distribution of biomolecular sequence combinations. This first probability distribution not only reflects the interaction possibility of each individual structural unit group (i.e., the first probability), but also provides information on which structural unit groups in the biomolecular sequence combination are more likely to form stable complexes.

[0122] Step 406 : Determine at least one first structural unit group from the plurality of candidate structural unit groups according to the first probability distribution.

[0123] Step 407 : predicting a target structure of the biomolecule complex based on the interacting structural units in each first structural unit group.

[0124] It should be noted that the execution process of step 401 and step 406 to step 407 can be implemented in any way in the embodiments of the present disclosure, and the embodiments of the present disclosure do not limit this and will not be described in detail.

[0125] In summary, by calculating the spatial distance between multiple structural units in any candidate structural unit group and determining the first probability of any candidate structural unit group based on the spatial distance, since the first probability is negatively correlated with the spatial distance, a smaller spatial distance corresponds to a higher first probability, which effectively quantifies the possibility of interaction between structural units in each candidate structural unit group; further, based on the first probability of each candidate structural unit group, a first probability distribution of biomolecular sequence combinations is generated. This first probability distribution provides a basis for screening high-potential first structural unit groups, thereby improving the reliability and accuracy of structural prediction of biomolecular complexes.

[0126] The above are embodiments corresponding to the application method of the structure prediction model (ie, structure prediction of a complex). The present disclosure also proposes a training method for the structure prediction model.

[0127] Figure 5 This is a flowchart of the training method of the structure prediction model provided in the fifth embodiment of the present disclosure.

[0128] like Figure 5 As shown, the training method of the structure prediction model may include the following steps:

[0129] Step 501: Obtain training samples.

[0130] The training samples include sample biomolecule sequence combinations, and the sample biomolecule sequence combinations are obtained based on a combination of multiple sample biomolecule sequences.

[0131] As one possible implementation, multiple sample biomolecule sequences are collected from a database, and sequence combinations are performed on these sample biomolecule sequences to obtain sample biomolecule sequence combinations. Training samples are generated based on these sample biomolecule sequence combinations. It should be noted that the training samples also include the annotated structures of the complexes corresponding to each sample biomolecule sequence combination.

[0132] Step 502 : Using a structure prediction model to predict a second probability distribution of sample biomolecule sequence combinations, and generating a predicted structure of a biomolecule complex based on the second probability distribution.

[0133] The second probability distribution is used to indicate a second probability of a plurality of candidate structural unit groups in the sample biomolecule sequence combination, wherein the candidate structural unit groups include structural units in at least two sample biomolecule sequences, and the second probability is used to indicate a possibility of interaction between the structural units in the candidate structural unit groups.

[0134] As a possible implementation method, the predicted sample biomolecule sequence combination in the training sample is input into a structure prediction model, and the structure prediction model predicts a second probability distribution of the sample biomolecule sequence combination, and generates a predicted structure of the biomolecule complex based on the second probability distribution; wherein the second probability distribution is used to indicate the second probability of multiple candidate structural unit groups in the sample biomolecule sequence combination, the candidate structural unit groups including structural units in at least two sample biomolecule sequences, and the second probability is used to indicate the possibility of interaction between each structural unit in the candidate structural unit group.

[0135] Step 503 : Training a structure prediction model based on the differences between the annotated structures and the predicted structures corresponding to the plurality of sample biomolecule sequences.

[0136] In an embodiment of the present disclosure, the value of the loss function (referred to as the loss value in the present disclosure) can be determined based on the difference between the labeled structure and the predicted structure, so that the structure prediction model can be trained according to the loss value to minimize the loss value.

[0137] It should be noted that the above only uses the termination condition of model training as the minimization of the loss value as an example. In actual application, other termination conditions can also be set. For example, the termination conditions can also include: the training time reaches the set time, the number of training times reaches the set number of times, etc. The embodiment of the present disclosure does not limit this.

[0138] In order to improve the accuracy and efficiency of the structure prediction of biomolecular complexes, in any embodiment of the present disclosure, a structure prediction model including a prediction network and a generation network is used as an example. The prediction network is used to predict the second probability distribution of the sample biomolecular sequence combination; the generation network is used to determine at least one second structural unit group from multiple candidate structural unit groups based on the second probability distribution; and the generation network is used to generate a predicted structure of the biomolecular complex based on the interacting structural units in the at least one second structural unit group.

[0139] In order to improve the prediction accuracy of the structure prediction model, in any embodiment of the present disclosure, a generative network is used to perform at least one round of prediction process on the interacting structural units in at least one second structural unit group; wherein the prediction process includes: generating a predicted structure of the biomolecule complex based on the interacting structural units in the retained second structural unit group; wherein the retained second structural unit group is obtained by filtering at least one second structural unit group based on the similarity between any two second structural unit groups.

[0140] In any embodiment of the present disclosure, the structure prediction model further includes a structure scoring network. The method of generating a predicted structure of a biomolecular complex based on interacting structural units in the retained second structural unit group mainly includes the following steps:

[0141] 1. Using a generative network, predicting candidate structures of biomolecular complexes corresponding to each retained second structural unit group based on the interacting structural units in each retained second structural unit group;

[0142] As an example, multiple rounds of sampling process are performed on the retained second structural unit group; wherein the sampling process includes: sampling an unlabeled second structural unit group from the retained second structural unit group, and labeling a second structural unit group; wherein the label is used to indicate that the second structural unit group has been sampled; and generating a candidate structure based on the interacting structural units in a second structural unit group.

[0143] 2. using a structure scoring network to predict a structure score for a candidate structure of the biomolecular complex corresponding to each retained second structural unit group; wherein the structure score indicates the degree of match between the candidate structure and a structural unit of the corresponding second structural unit group;

[0144] 3. Determine a target structure from the candidate structures of the biomolecular complex corresponding to each retained second structural unit group based on the structure scores of the candidate structures of the biomolecular complex corresponding to each retained second structural unit group.

[0145] The structure prediction model training method of the disclosed embodiments obtains training samples comprising sample biomolecular sequence combinations, wherein the sample biomolecular sequence combinations are generated from multiple sample biomolecular sequences to provide rich input data for the structure prediction model. The model is used to predict a second probability distribution of these sample biomolecular sequence combinations, and based on this probability distribution, a predicted structure of a biomolecular complex is generated. The second probability distribution quantifies the likelihood of interaction between structural units in each candidate structural unit group. Based on the second probability distribution, the structural unit groups most likely to interact can be screened to further predict the three-dimensional structure of the biomolecular complex, avoiding a blind search of a massive conformational space. The method also ensures that the structure of the generated biomolecular complex meets its interaction requirements by using the interacting structural units in the structural unit group as a guide. Finally, the structure prediction model is optimized and trained based on the differences between the known annotated structures corresponding to the multiple sample biomolecular sequences and the predicted structures generated by the model, thereby improving the accuracy and reliability of the model in predicting the structure of the biomolecular complex.

[0146] Based on any of the above embodiments, Figure 6 As shown, the structure prediction model is an all-atom structure prediction model, which includes: a prediction network (also called a contact prediction network), a generation network (also called a structure generation network with contact guidance), and a structure scoring network (also called a structure scoring network) for example. The structure prediction method of the complex of the embodiment of the present disclosure can also be implemented based on the following steps:

[0147] (1) obtaining a biomolecule sequence combination, wherein the biomolecule sequence combination includes a biomolecule sequence A (A1-A2-A3-A4) and a biomolecule sequence B (B1-B2-B3);

[0148] (2) Using a prediction network to predict the probability distribution of biomolecular sequence combinations (also known as contact probability distribution), based on this probability distribution, a set of contacts with high probability (the first structural unit group) is identified;

[0149] (3) Sampling is performed based on the retained high-probability contact set (retained first structural unit group) to generate conformations that can satisfy the contact. Since these contacts may correspond to similar binding modes, the contact sets with similar binding modes are grouped, that is, the high-probability contact sets are filtered to obtain the retained high-probability contact set. This method includes the following steps:

[0150] 1) sampling a contact from the retained high-probability contact set (i.e., sampling a retained first structural unit group);

[0151] 2) Generate a network for structure prediction. The model will generate a conformation that conforms to the input contact.

[0152] 3) Resample a contact that has not been sampled yet and repeat the previous steps;

[0153] 4) If all high-probability contact sets have been sampled and the sampling process is still ongoing, resample all high-probability contact sets;

[0154] Finally, all predicted structures are ranked according to the scores given by the structure scoring network, and the structure with the highest score is selected as the final prediction result.

[0155] With the above Figures 1 to 4 Corresponding to the structure prediction method of the complex provided in the embodiment, the present disclosure also provides a structure prediction device for the complex. Figures 1 to 4 The structure prediction method of the complex provided in the embodiment corresponds to the embodiment, so the implementation of the structure prediction method of the complex is also applicable to the structure prediction device of the complex provided in the embodiment of the present disclosure, and will not be described in detail in the embodiment of the present disclosure.

[0156] Figure 7 This is a schematic diagram of the structure of the complex structure prediction device provided in Example 6 of the present disclosure.

[0157] like Figure 7 As shown, the structure prediction device 700 for a complex includes: an acquisition module 710 , a prediction module 720 , a determination module 730 and a generation module 740 .

[0158] Among them, the acquisition module 710 is used to obtain a biomolecule sequence combination; wherein the biomolecule sequence combination is obtained by sequence combination based on a specified plurality of biomolecule sequences; the prediction module 720 is used to predict a first probability distribution of the biomolecule sequence combination; wherein the probability distribution is used to indicate a first probability of a plurality of candidate structural unit groups in the biomolecule sequence combination, wherein the candidate structural unit group includes structural units from at least two biomolecule sequences, and the probability is used to indicate the possibility of interaction between the structural units in the candidate structural unit group; the determination module 730 is used to determine at least one first structural unit group from the plurality of candidate structural unit groups based on the first probability distribution; and the generation module 740 is used to predict a target structure of the biomolecule complex based on the interacting structural units in each first structural unit group.

[0159] As a possible implementation, the generation module 740 is configured to obtain the similarity between any two first structural unit groups; wherein the similarity is used to indicate the degree of similarity between biomolecule complexes generated based on the interacting structural units in any two first structural unit groups; based on the similarity between each first structural unit group, each first structural unit group is filtered to obtain a retained target structural unit group; and based on the interacting structural units in the retained first structural unit group, a target structure of the biomolecule complex is predicted.

[0160] As a possible implementation method, a generation module 740 is used to predict candidate structures of the biomolecular complex corresponding to each retained first structural unit group based on the interacting structural units in each retained first structural unit group; predict a structural score for each candidate structure; wherein the structural score is used to indicate the degree of match between the candidate structure and the structural unit of the corresponding first structural unit group; and determine a target structure from each candidate structure based on the structural score of each candidate structure.

[0161] As a possible implementation method, a generation module 740 is used to perform multiple rounds of sampling processes on the retained first structural unit group; wherein the sampling process includes: sampling an unlabeled first structural unit group from the retained first structural unit group, and labeling a first structural unit group; wherein the label is used to indicate that the first structural unit group has been sampled; and generating a candidate structure based on the interacting structural units in a first structural unit group.

[0162] As a possible implementation, generation module 740 is configured to perform structure prediction based on any two first structural unit groups to obtain predicted structures of any two first structural unit groups; wherein the predicted structures are used to indicate the structures of biomolecule complexes generated based on the first structural unit groups; based on the predicted structures of any two first structural unit groups, evaluate the similarity between the structures of the biomolecule complexes generated by any two first structural unit groups; and determine the similarity between any two first structural unit groups based on the similarity.

[0163] As a possible implementation, the determination module 730 is configured to determine at least one first structural unit group from each candidate structural unit group based on the first probability of each candidate structural unit group in the first probability distribution; wherein the first probability of the first structural unit group is greater than a set threshold.

[0164] As a possible implementation, the prediction module 720 is configured to obtain multiple candidate structural unit groups in a biomolecule sequence combination; calculate the spatial distance between multiple structural units in any candidate structural unit group; determine a first probability of any candidate structural unit group based on the spatial distance between multiple structural units in any candidate structural unit group, wherein the probability is negatively correlated with the spatial distance; and generate a first probability distribution of the biomolecule sequence combination based on the first probability of each candidate structural unit group.

[0165] The structure prediction device for the complex of the embodiment of the present disclosure can simulate the potential interaction patterns between different biomolecules by obtaining a biomolecule sequence combination composed of multiple specified biomolecule sequences, thereby providing a rich candidate space for subsequent structure prediction; further, based on the first probability distribution of the predicted biomolecule sequence combination, the possibility of interaction between structural units in each candidate structural unit group in the biomolecule sequence combination can be quantitatively evaluated, thereby screening out the first structural unit group with the greatest possibility of interaction, avoiding blind search of massive conformational space; finally, based on the information of the interacting structural units in the first structural unit group, the overall three-dimensional structure of the target biomolecule complex is further predicted, and the interacting structural units in the first structural unit group are used as a guide to ensure that the structure of the generated biomolecule complex meets its interaction requirements, thereby improving the reliability and biological rationality of the structure prediction.

[0166] With the above Figures 5 and 6 Corresponding to the training method of the structure prediction model provided in the embodiment, the present disclosure also provides a training device for the structure prediction model. Figures 5 and 6 The training method of the structure prediction model provided in the embodiment corresponds to the embodiment, so the implementation method of the structure prediction model training method is also applicable to the training device of the structure prediction model provided in the embodiment of the present disclosure, and will not be described in detail in the embodiment of the present disclosure.

[0167] Figure 8 This is a structural diagram of the training device for the structure prediction model provided in Example 7 of the present disclosure.

[0168] like Figure 8 As shown, the training device 800 for the structure prediction model may include: an acquisition module 810 , a prediction module 820 and a training module 830 .

[0169] Among them, the acquisition module 810 is used to obtain training samples; wherein the training samples include sample biomolecule sequence combinations, and the sample biomolecule sequence combinations are obtained based on multiple sample biomolecule sequence combinations; the prediction module 820 is used to use the structure prediction model to predict a second probability distribution of the sample biomolecule sequence combinations, and generate a predicted structure of the biomolecule complex based on the second probability distribution; wherein the second probability distribution is used to indicate the second probability of multiple candidate structural unit groups in the sample biomolecule sequence combinations, the candidate structural unit groups including structural units in at least two sample biomolecule sequences, and the second probability is used to indicate the possibility of interaction between each structural unit in the candidate structural unit group; and the training module 830 is used to train the structure prediction model based on the differences between the annotated structures corresponding to the multiple sample biomolecule sequences and the predicted structures.

[0170] As a possible implementation, the structure prediction model includes a prediction network and a generation network, and the prediction module 820 is used to use the prediction network to predict a second probability distribution of the sample biological molecule sequence combination; use the generation network to determine at least one second structural unit group from multiple candidate structural unit groups based on the second probability distribution; and use the generation network to generate a predicted structure of the biological molecule complex based on the interacting structural units in the at least one second structural unit group.

[0171] As a possible implementation manner, the prediction module is used to use the generation network to perform at least one round of prediction process on the interacting structural units in at least one second structural unit group; wherein the prediction process includes: generating a predicted structure of the biomolecule complex based on the interacting structural units in the retained second structural unit group; wherein the retained second structural unit group is obtained by filtering at least one second structural unit group based on the similarity between any two second structural unit groups.

[0172] The training apparatus for a structure prediction model according to an embodiment of the present disclosure obtains training samples comprising sample biomolecular sequence combinations, wherein the sample biomolecular sequence combinations are generated from multiple sample biomolecular sequences to provide rich input data for a structure prediction model. The model is then used to predict a second probability distribution for these sample biomolecular sequence combinations, and a predicted structure of a biomolecular complex is generated based on this probability distribution. The second probability distribution quantifies the likelihood of interaction between structural units in each candidate structural unit group. Based on the second probability distribution, the structural unit groups most likely to interact can be screened for further prediction of the three-dimensional structure of the biomolecular complex, thereby avoiding blind searching of a vast conformational space and ensuring that the structure of the generated biomolecular complex meets its interaction requirements using the interacting structural units in the structural unit group as a guide. Finally, the structure prediction model is optimized and trained based on the differences between the known annotated structures corresponding to the multiple sample biomolecular sequences and the predicted structures generated by the model, thereby improving the accuracy and reliability of the model in predicting the structure of the biomolecular complex.

[0173] In order to implement the above embodiments, the present disclosure also provides an electronic device, which may include at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the structure prediction method of the complex proposed in any of the above embodiments of the present disclosure, or to execute the training method of the structure prediction model proposed in any of the above embodiments of the present disclosure.

[0174] In order to implement the above embodiments, the present disclosure also provides a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to enable a computer to execute the structure prediction method of the complex proposed in any of the above embodiments of the present disclosure, or to execute the training method of the structure prediction model proposed in any of the above embodiments of the present disclosure.

[0175] In order to implement the above embodiments, the present disclosure also provides a computer program product, which includes a computer program. When the computer program is executed by a processor, it implements the structure prediction method of the complex proposed in any of the above embodiments of the present disclosure, or implements the training method of the structure prediction model proposed in any of the above embodiments of the present disclosure.

[0176] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0177] Figure 9A schematic block diagram of an example electronic device that can be used to implement an embodiment of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0178] like Figure 9 As shown, the electronic device 900 includes a computing unit 901, which can perform various appropriate actions and processes according to a computer program stored in a ROM (Read-Only Memory) 902 or a computer program loaded from a storage unit 908 into a RAM (Random Access Memory) 903. In the RAM 903, various programs and data required for the operation of the electronic device 900 can also be stored. The computing unit 901, the ROM 902, and the RAM 903 are connected to each other via a bus 904. An I / O (Input / Output) interface 905 is also connected to the bus 904.

[0179] Multiple components in the electronic device 900 are connected to the I / O interface 905, including an input unit 906, such as a keyboard, a mouse, etc.; an output unit 907, such as various types of displays, speakers, etc.; a storage unit 908, such as a magnetic disk, an optical disk, etc.; and a communication unit 909, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 909 allows the electronic device 900 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0180] The computing unit 901 can be a variety of general-purpose and / or specialized processing components with processing and computing capabilities. Some examples of the computing unit 901 include, but are not limited to, a CPU (Central Processing Unit), a GPU (Graphic Processing Units), various dedicated AI (Artificial Intelligence) computing chips, various computing units that run machine learning model algorithms, a DSP (Digital Signal Processor), and any appropriate processor, controller, microcontroller, etc. The computing unit 901 performs the various methods and processes described above, such as the structure prediction method of the above-mentioned complex or the training method of the molecular structure prediction model. For example, in some embodiments, the structure prediction method of the above-mentioned complex or the training method of the structure prediction model can be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as the storage unit 908. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 900 via the ROM 902 and / or the communication unit 909. When the computer program is loaded into RAM 903 and executed by the computing unit 901, one or more steps of the above-described method for generating a bicyclic peptide or the method for training a molecular structure prediction model can be performed. Alternatively, in other embodiments, the computing unit 901 can be configured to perform the above-described method for predicting the structure of the complex or the method for training a structure prediction model by any other appropriate means (e.g., by means of firmware).

[0181] Various embodiments of the systems and techniques described herein can be implemented in digital electronic circuit systems, integrated circuit systems, FPGAs (Field Programmable Gate Arrays), ASICs (Application-Specific Integrated Circuits), ASSPs (Application-Specific Standard Products), SOCs (System on Chips), CPLDs (Complex Programmable Logic Devices), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include being implemented in one or more computer programs that are executable and / or interpreted on a programmable system that includes at least one programmable processor, which can be a special-purpose or general-purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0182] The program code for implementing the method of the present disclosure can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device so that when the program code is executed by the processor or controller, the functions / operations specified in the flow chart and / or block diagram are implemented. The program code can be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0183] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or apparatus. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or apparatus, or any suitable combination of the foregoing. More specific examples of machine-readable storage media may include an electrical connection based on one or more wires, a portable computer disk, a hard disk, RAM, ROM, EPROM (Electrically Programmable Read-Only-Memory) or flash memory, optical fiber, CD-ROM (Compact Disc Read-Only Memory), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0184] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (Cathode-Ray Tube) or LCD (Liquid Crystal Display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0185] The systems and techniques described herein can be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: LAN (Local Area Network), WAN (Wide Area Network), the Internet, and blockchain networks.

[0186] A computer system may include a client and a server. The client and server are generally remote from each other and typically interact via a communication network. The client-server relationship is established by computer programs running on the respective computers and establishing a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, a host product within the cloud computing service ecosystem that addresses the management difficulties and poor scalability of traditional physical hosts and VPS services. The server may also be a server in a distributed system or a server integrated with blockchain.

[0187] It's important to note that artificial intelligence (AI) is the study of how computers can simulate certain human thought processes and intelligent behaviors (such as learning, reasoning, thinking, and planning). This encompasses both hardware and software technologies. AI hardware technologies generally include sensors, specialized AI chips, cloud computing, distributed storage, and big data processing. AI software technologies primarily encompass computer vision, speech recognition, natural language processing, machine learning / deep learning, big data processing, and knowledge graphs.

[0188] According to the technical solution of the embodiment of the present disclosure, by obtaining a biomolecular sequence combination composed of a plurality of specified biomolecular sequences, the potential interaction patterns between different biomolecules can be simulated, thereby providing a rich candidate space for subsequent structure prediction; further, based on the first probability distribution of the predicted biomolecular sequence combination, the possibility of interaction between the structural units in each candidate structural unit group in the biomolecular sequence combination is quantitatively evaluated, thereby screening out the first structural unit group with the greatest possibility of interaction, thereby avoiding blind searching of the massive conformational space; finally, based on the information of the interacting structural units in the first structural unit group, the overall three-dimensional structure of the target biomolecular complex is further predicted, thereby achieving the goal of using the interacting structural units in the first structural unit group as a guide to ensure that the structure of the generated biomolecular complex meets its interaction requirements, thereby improving the reliability and biological rationality of the structure prediction.

[0189] It should be understood that the various forms of the processes shown above can be used to reorder, add, or delete steps. For example, the steps described in this disclosure can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions proposed in this disclosure can be achieved. This is not limited herein.

[0190] The above specific embodiments do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure shall be included within the scope of protection of this disclosure.

Claims

1. A method for predicting the structure of a complex, comprising: Obtaining a biomolecule sequence combination; wherein the biomolecule sequence combination is obtained by sequence combination based on a plurality of specified biomolecule sequences; Predicting a first probability distribution of the biomolecule sequence combination; wherein the first probability distribution is used to indicate first probabilities of a plurality of candidate structural unit groups in the biomolecule sequence combination, the candidate structural unit groups comprising structural units from at least two of the biomolecule sequences, and the first probabilities are used to indicate likelihoods of interactions between structural units in the candidate structural unit groups; determining at least one first structural unit group from the plurality of candidate structural unit groups according to the first probability distribution; A target structure of the biomolecule complex is predicted based on the interacting structural units in each of the first structural unit groups.

2. The method according to claim 1, wherein The step of predicting a target structure of a biomolecule complex based on the interacting structural units in each of the first structural unit groups comprises: Obtaining a similarity between any two first structural unit groups; wherein the similarity is used to indicate a degree of similarity between biomolecule complexes generated based on interacting structural units in any two first structural unit groups; filtering the first structural unit groups based on the similarities between the first structural unit groups to obtain retained first structural unit groups; A target structure of the biomolecular complex is predicted based on the interacting structural units in the retained first structural unit group.

3. The method according to claim 2, wherein: The step of predicting a target structure of a biomolecule complex based on the interacting structural units in the retained first structural unit group comprises: predicting candidate structures of biomolecular complexes corresponding to each of the retained first building block groups based on the interacting building blocks in each of the retained first building block groups; Predicting a structural score for each of the candidate structures; wherein the structural score is used to indicate the degree of matching between the candidate structure and the corresponding structural unit of the first structural unit group; A target structure is determined from the candidate structures according to the structure scores of the candidate structures.

4. The method according to claim 3, wherein: The step of predicting candidate structures of biomolecular complexes corresponding to each of the retained first structural unit groups based on the interacting structural units in each of the retained first structural unit groups comprises: Performing multiple rounds of sampling on the retained first structural unit group; wherein the sampling process includes: Sampling an unmarked first structural unit group from the retained first structural unit group, and marking the first structural unit group; wherein the mark is used to indicate that the first structural unit group has been sampled; A candidate structure is generated according to the interacting structural units in the first structural unit group.

5. The method according to claim 2, wherein: The obtaining of the similarity between any two first structural unit groups includes: Performing structure prediction based on any two first structural unit groups to obtain predicted structures of the any two first structural unit groups; wherein the predicted structures are used to indicate the structure of a biomolecule complex generated based on the first structural unit groups; Based on the predicted structures of any two first building block groups, evaluating the similarity between the structures of the biomolecular complexes generated by any two first building block groups; The similarity between any two first structural unit groups is determined according to the similarity degree.

6. The method according to claim 1, wherein Determining at least one first structural unit group from the plurality of candidate structural unit groups according to the first probability distribution includes: determining at least one first structural unit group from each of the candidate structural unit groups according to the first probability of each of the candidate structural unit groups in the first probability distribution; The first probability of the first structural unit group is greater than a set threshold.

7. The method according to claim 1, wherein The predicting a first probability distribution of the biomolecule sequence combination includes: obtaining a plurality of candidate structural unit groups in the biomolecule sequence combination; Calculating the spatial distances between multiple structural units in any candidate structural unit group; Determining a first probability of any candidate structural unit group according to the spatial distance between multiple structural units in any candidate structural unit group; wherein the first probability is negatively correlated with the spatial distance; Based on the first probabilities of the candidate building block groups, a first probability distribution of the biomolecule sequence combinations is generated.

8. A method for training a structure prediction model, comprising: Acquire a training sample; wherein the training sample includes a sample biomolecule sequence combination, and the sample biomolecule sequence combination is obtained based on a plurality of sample biomolecule sequence combinations; A structure prediction model is used to predict a second probability distribution of the sample biomolecule sequence combination, and a predicted structure of the biomolecule complex is generated based on the second probability distribution; wherein the second probability distribution is used to indicate second probabilities of a plurality of candidate structural unit groups in the sample biomolecule sequence combination, the candidate structural unit groups including structural units in at least two of the sample biomolecule sequences, and the second probabilities are used to indicate the likelihood of interaction between the structural units in the candidate structural unit groups; The structure prediction model is trained based on the differences between the annotated structures corresponding to the plurality of sample biomolecule sequences and the predicted structures.

9. The method according to claim 8, wherein The structure prediction model includes a prediction network and a generation network, and the predicted biomolecule complex is generated by the structure prediction model using the following steps: Using the prediction network, predicting a second probability distribution of the sample biomolecule sequence combination; Determining at least one second structural unit group from the plurality of to-be-selected structural unit groups using the generation network according to the second probability distribution; as well as The generated network is used to generate a predicted structure of the biomolecular complex based on the interacting building blocks in the at least one second set of building blocks.

10. The method according to claim 9, wherein: The generating network is used to generate a predicted structure of the biomolecular complex based on the interacting structural units in the at least one second structural unit group, comprising: Using the generative network, at least one round of prediction process is performed on the interacting structural units in at least one second structural unit group; wherein the prediction process includes: Based on the interacting structural units in the retained second structural unit group, a predicted structure of the biomolecule complex is generated; wherein the retained second structural unit group is obtained by filtering the at least one second structural unit group based on the similarity between any two second structural unit groups.

11. A structure prediction device for a complex, comprising: An acquisition module, configured to acquire a biomolecule sequence combination; wherein the biomolecule sequence combination is obtained by combining a plurality of specified biomolecule sequences; a prediction module configured to predict a first probability distribution of the biomolecule sequence combination; wherein the probability distribution indicates first probabilities of a plurality of candidate structural unit groups in the biomolecule sequence combination, the candidate structural unit groups comprising structural units from at least two of the biomolecule sequences, and the probabilities indicating likelihood of interaction between structural units in the candidate structural unit groups; a determining module, configured to determine at least one first structural unit group from the plurality of candidate structural unit groups according to the first probability distribution; A generating module is used to predict a target structure of a biomolecule complex based on the interacting structural units in each of the first structural unit groups.

12. A training device for a structure prediction model, comprising: An acquisition module, configured to acquire a training sample; wherein the training sample comprises a sample biomolecule sequence combination, and the sample biomolecule sequence combination is obtained based on a plurality of sample biomolecule sequence combinations; a prediction module configured to predict a second probability distribution of the sample biomolecule sequence combination using a structure prediction model, and generate a predicted structure of the biomolecule complex based on the second probability distribution; wherein the second probability distribution is configured to indicate a second probability of a plurality of candidate structural unit groups in the sample biomolecule sequence combination, wherein the candidate structural unit groups include structural units from at least two of the sample biomolecule sequences, and the second probability is configured to indicate a likelihood of interaction between the structural units in the candidate structural unit groups; A training module is used to train the structure prediction model according to the differences between the annotated structures corresponding to the multiple sample biomolecule sequences and the predicted structures.

13. An electronic device, wherein: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the method of any one of claims 1 to 7, or the method of any one of claims 8 to 10.

14. A non-transitory computer-readable storage medium storing computer instructions, wherein: The computer instructions are used to enable the computer to execute the method according to any one of claims 1 to 7, or to execute the method according to any one of claims 8 to 10.

15. A computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the computer program implements the steps of the method according to any one of claims 1 to 7, or implements the steps of the method according to any one of claims 8 to 10.

Images