State prediction method, device, apparatus and storage medium

Abstract

The application discloses a state prediction method and device, equipment and a storage medium, and belongs to the computer technical field. Through the technical scheme provided by the embodiment of the application, the state prediction model extracts the gene information and sequence of the immune cell receptor to obtain the gene features and sequence features of the immune cell receptor. In the process of obtaining the receptor features of the immune cell receptor, the gene features, sequence features and three-dimensional structure features are fused. The introduction of the three-dimensional structure features enriches the content of the receptor features and improves the expression ability of the receptor features, so that the accuracy of the state prediction of the target object based on the receptor features is higher.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of computer, in particular to a state prediction method and device, equipment and storage medium. BACKGROUND

[0002] The human immune system is composed of innate immunity and adaptive immunity. The adaptive immune system is implemented by a variety of immune cells, which can specifically respond to specific pathogens. Immune cell receptors are the regions of immune cells that recognize antigens, and successful recognition of antigens can activate the immune system to eliminate pathogens, playing an important role in maintaining human health. The immune repertoire is the sum of all immune cells in the circulatory system of a certain individual, and studying the relationship between the immune repertoire and the state (such as disease) helps to explore tumor immune mechanisms, discover disease treatment targets, develop antibodies and evaluate vaccine effectiveness. Therefore, there is an urgent need for a method for predicting the state of an individual based on the immune repertoire. SUMMARY

[0003] The embodiments of the present application provide a state prediction method, device, equipment and storage medium, which can predict the state of an individual based on the immune repertoire. The technical solution is as follows:

[0004] In one aspect, a state prediction method is provided, which includes:

[0005] Inputting the gene information, sequence information and three-dimensional structure characteristics of a plurality of immune cell receptors in the immune repertoire of a target object into a state prediction model;

[0006] Extracting the gene information and sequence information of the plurality of immune cell receptors through the state prediction model to obtain the gene characteristics and sequence characteristics of each immune cell receptor;

[0007] Fusing the gene characteristics, sequence characteristics and three-dimensional structure characteristics of each immune cell receptor through the state prediction model to obtain the receptor characteristics of each immune cell receptor;

[0008] Performing state prediction based on the receptor characteristics of each immune cell receptor through the state prediction model, and outputting at least one target state of the target object.

[0009] In one aspect, a training method of a state prediction model is provided, which includes:

[0010] Inputting the gene information, sequence information and three-dimensional structure characteristics of a plurality of immune cell receptors in the immune repertoire of a sample object into a state prediction model;

[0011] The state prediction model is used for feature extraction on the gene information and the sequence information of the plurality of immune cell receptors, to obtain gene features and sequence features of each immune cell receptor;

[0012] The state prediction model is used for fusion of the gene features, the sequence features and the three-dimensional structure features of each immune cell receptor, to obtain receptor features of each immune cell receptor;

[0013] The state prediction model is used for state prediction based on the receptor features of each immune cell receptor, to output at least one sample state of the sample object;

[0014] The state prediction model is trained based on difference information between the at least one sample state of the sample object and a labeled state.

[0015] In one aspect, a state prediction apparatus is provided, and the apparatus comprises:

[0016] An input unit is configured to input gene information, sequence information and three-dimensional structure features of a plurality of immune cell receptors in an immune repertoire of a target object into a state prediction model;

[0017] A feature extraction unit is configured to use the state prediction model to perform feature extraction on the gene information and the sequence information of the plurality of immune cell receptors, to obtain gene features and sequence features of each immune cell receptor;

[0018] A feature fusion unit is configured to use the state prediction model to perform fusion of the gene features, the sequence features and the three-dimensional structure features of each immune cell receptor, to obtain receptor features of each immune cell receptor;

[0019] A state prediction unit is configured to use the state prediction model to perform state prediction based on the receptor features of each immune cell receptor, to output at least one target state of the target object.

[0020] In one possible implementation, the feature extraction unit is configured to, for any immune cell receptor in the plurality of immune cell receptors, use a gene encoder of the state prediction model to encode VDJ information of the immune cell receptor, to obtain gene features of the immune cell receptor, where V represents a variable region, D represents a hypervariable region, and J represents a junction region; and use a sequence encoder of the state prediction model to encode an amino acid sequence of the immune cell receptor, to obtain sequence features of the immune cell receptor.

[0021] In one possible implementation, the feature extraction unit is configured to perform any one of the following:

[0022] In the case where the immune cell receptor is a B cell receptor, the VJ information of the light chain and the VDJ information of the heavy chain of the immune cell receptor are encoded to obtain the genetic feature of the immune cell receptor;

[0023] In the case where the immune cell receptor is a T cell receptor, the VJ information of the alpha chain and the VDJ information of the beta chain of the immune cell receptor are encoded to obtain the genetic feature of the immune cell receptor.

[0024] In a possible implementation, the feature extraction unit is configured to perform full connection on the VJ information of the light chain and the VDJ information of the heavy chain of the immune cell receptor to obtain the genetic feature of the immune cell receptor, the genetic feature of the immune cell receptor including a light chain genetic feature of the immune cell receptor and a heavy chain genetic feature of the immune cell receptor; and the encoding of the VJ information of the alpha chain and the VDJ information of the beta chain of the immune cell receptor to obtain the genetic feature of the immune cell receptor includes full connection on the VJ information of the alpha chain and the VDJ information of the beta chain of the immune cell receptor to obtain the genetic feature of the immune cell receptor, the genetic feature of the immune cell receptor including an alpha chain genetic feature of the immune cell receptor and a beta chain genetic feature of the immune cell receptor.

[0025] In a possible implementation, the feature extraction unit is configured to perform any one of the following:

[0026] In the case where the immune cell receptor is a B cell receptor, the amino acid sequence of the light chain and the amino acid sequence of the heavy chain of the immune cell receptor are encoded based on an attention mechanism by a sequence encoder of the state prediction model to obtain a sequence feature of the immune cell receptor, the sequence feature of the immune cell receptor including a light chain sequence feature of the immune cell receptor and a heavy chain sequence feature of the immune cell receptor;

[0027] In the case where the immune cell receptor is a T cell receptor, the amino acid sequence of the alpha chain and the amino acid sequence of the beta chain of the immune cell receptor are encoded based on an attention mechanism by a sequence encoder of the state prediction model to obtain a sequence feature of the immune cell receptor, the sequence feature of the immune cell receptor including an alpha chain sequence feature of the immune cell receptor and a beta chain sequence feature of the immune cell receptor.

[0028] In a possible implementation, the feature fusion unit is configured to, for any immune cell receptor in the plurality of immune cell receptors, splice the genetic feature and the sequence feature of the immune cell receptor by a feature fusioner of the state prediction model to obtain a genetic sequence fusion feature of the immune cell receptor;

[0029] Based on the gating attention mechanism, the gene sequence fusion features and three-dimensional structural features of the immune cell receptor are weighted and fused to obtain the receptor features of the immune cell receptor.

[0030] In one possible implementation, the state prediction unit is configured to perform fully connected and normalized receptor features of each immune cell receptor through the classification module of the state prediction model, and output the probability of each immune cell receptor corresponding to multiple candidate states; and determine at least one target state of the target object based on the probability of each immune cell receptor corresponding to multiple candidate states.

[0031] In one possible implementation, the state prediction unit is configured to determine a reference candidate state corresponding to each immune cell receptor based on the probability of each immune cell receptor corresponding to a plurality of candidate states, wherein the reference candidate state is a candidate state whose probability meets a first probability condition among the plurality of candidate states; and to determine at least one target state of the target object based on the reference candidate states corresponding to the plurality of immune cell receptors.

[0032] In one possible implementation, the state prediction unit is configured to perform any of the following:

[0033] For the first reference candidate state among the reference candidate states corresponding to the plurality of immune cell receptors, if the number of times the plurality of immune cell receptors correspond to the first reference candidate state meets the number condition, the first reference candidate state is determined as the target state of the target object.

[0034] For the second reference candidate state among the reference candidate states corresponding to the plurality of immune cell receptors, determine the average probability of the plurality of immune cell receptors corresponding to the second reference candidate state; if the average probability meets the second probability condition, determine the second reference candidate state as the target state of the target object.

[0035] In one possible implementation, the state prediction unit is configured to perform any of the following:

[0036] The reference candidate states corresponding to the plurality of immune cell receptors are sorted in order of probability from largest to smallest; the top number of reference candidate states are determined as the target state of the target object.

[0037] The reference candidate states corresponding to the plurality of immune cell receptors and the probabilities corresponding to the reference candidate states are input into the state screening model; the state screening model processes the reference candidate states corresponding to the plurality of immune cell receptors and the probabilities corresponding to the reference candidate states, and outputs the at least one target state.

[0038] In a possible implementation, the apparatus further includes a three-dimensional structure feature determination unit configured to, for any immune cell receptor in the plurality of immune cell receptors, obtain an amino acid sequence of a CDR3 region of the immune cell receptor; perform multiple sequence alignment on the amino acid sequence of the CDR3 region of the immune cell receptor to obtain at least one reference amino acid sequence, similarity between the reference amino acid sequence and the amino acid sequence of the CDR3 region of the immune cell receptor meeting a similarity condition; obtain a homologous template corresponding to the amino acid sequence of the CDR3 region of the immune cell receptor, the homologous template including structural information of a homologous sequence of the amino acid sequence of the CDR3 region of the immune cell receptor; and perform multiple iterations based on the amino acid sequence of the CDR3 region of the immune cell receptor, the at least one reference amino acid sequence, and the homologous template to obtain a three-dimensional structure feature of the immune cell receptor.

[0039] In a possible implementation, the apparatus further includes a three-dimensional structure feature determination unit configured to, for any immune cell receptor in the plurality of immune cell receptors, obtain three-dimensional structure information of the immune cell receptor, the three-dimensional structure information including three-dimensional coordinates of a plurality of amino acids in the immune cell receptor.

[0040] The three-dimensional structure feature determination unit is further configured to perform any one of the following:

[0041] perform graph convolution on the three-dimensional structure information of the immune cell receptor to obtain a three-dimensional structure feature of the immune cell receptor;

[0042] encode the three-dimensional structure information of the immune cell receptor based on an attention mechanism to obtain a three-dimensional structure feature of the immune cell receptor.

[0043] In a possible implementation, the feature fusion unit is further configured to fuse, by the state prediction model, the genetic feature, the sequence feature, the three-dimensional structure feature of each of the immune cell receptors, and the physicochemical information of the amino acids in each of the immune cell receptors to obtain a receptor feature of each of the immune cell receptors.

[0044] In an aspect, a training apparatus of a state prediction model is provided, and the apparatus includes:

[0045] a training information input unit configured to input genetic information, sequence information, and three-dimensional structure features of a plurality of immune cell receptors in an immune repertoire of a sample object into a state prediction model;

[0046] The training feature extraction unit is configured to perform feature extraction on the gene information and the sequence information of the plurality of immune cell receptors by using the state prediction model, to obtain gene features and sequence features of each of the immune cell receptors.

[0047] The training feature fusion unit is configured to fuse the gene features, the sequence features and the three-dimensional structure features of each of the immune cell receptors by using the state prediction model, to obtain receptor features of each of the immune cell receptors.

[0048] The sample state output unit is configured to perform state prediction based on the receptor features of each of the immune cell receptors by using the state prediction model, and output at least one sample state of the sample object.

[0049] The training unit is configured to train the state prediction model based on difference information between the at least one sample state and the labeled state of the sample object.

[0050] In an aspect, a computer device is provided, which includes one or more processors and one or more memories, and the one or more memories store at least one computer program, which is loaded and executed by the one or more processors to implement the state prediction method or the training method of the state prediction model.

[0051] In an aspect, a computer readable storage medium is provided, which stores at least one computer program, which is loaded and executed by a processor to implement the state prediction method or the training method of the state prediction model.

[0052] In an aspect, a computer program product or a computer program is provided, which includes program code stored in a computer readable storage medium, and a processor of a computer device reads the program code from the computer readable storage medium, and the processor executes the program code to enable the computer device to perform the state prediction method or the training method of the state prediction model.

[0053] By the technical solutions provided in the embodiments of the present application, the state prediction model performs feature extraction on the gene information and the sequence of the immune cell receptors, to obtain gene features and sequence features of the immune cell receptors. In the process of obtaining the receptor features of the immune cell receptors, the gene features, the sequence features and the three-dimensional structure features are fused. The introduction of the three-dimensional structure features enriches the content of the receptor features and improves the expression ability of the receptor features, so that the accuracy of state prediction of a target object based on the receptor features is relatively high. BRIEF DESCRIPTION OF DRAWINGS

[0054] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the embodiments description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without any creative effort on the basis of these drawings.

[0055] Figure 1 is a schematic diagram of an implementation environment of a state prediction method provided by an embodiment of the present application;

[0056] Figure 2 is a flowchart of a state prediction method provided by an embodiment of the present application;

[0057] Figure 3 is a flowchart of another state prediction method provided by an embodiment of the present application;

[0058] Figure 4 is a flowchart of a method for determining a three-dimensional structure feature provided by an embodiment of the present application;

[0059] Figure 5 is a flowchart of a method for determining a probability of a candidate state provided by an embodiment of the present application;

[0060] Figure 6 is a flowchart of another state prediction method provided by an embodiment of the present application;

[0061] Figure 7 is a schematic diagram of a state prediction interface provided by an embodiment of the present application;

[0062] Figure 8 is a schematic diagram of a state display interface provided by an embodiment of the present application;

[0063] Figure 9 is a schematic diagram of an experimental result provided by an embodiment of the present application;

[0064] Figure 10 is another schematic diagram of an experimental result provided by an embodiment of the present application;

[0065] Figure 11 is a flowchart of a training method of a state prediction model provided by an embodiment of the present application;

[0066] Figure 12 is a structural schematic diagram of a state prediction device provided by an embodiment of the present application;

[0067] Figure 13 is a structural schematic diagram of a training device of a state prediction model provided by an embodiment of the present application;

[0068] Figure 14is a structural schematic diagram of a terminal provided by an embodiment of the present application.

[0069] Figure 15 is a structural schematic diagram of a server provided by an embodiment of the present application. DETAILED DESCRIPTION

[0070] In order to make the purpose, technical solutions and advantages of the present application clearer, the embodiments of the present application will be further described in detail below with reference to the drawings.

[0071] In the present application, the terms "first", "second", and the like are used to distinguish the same or similar items with basically the same function and purpose, and it should be understood that there is no logical or time sequence dependency between "first", "second", and "nth", and the number and execution order are not limited.

[0072] Artificial intelligence (AI) is the use of digital computers or digital computer-controlled machines to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results. In other words, artificial intelligence is a comprehensive technology of computer science, which tries to understand the essence of intelligence and produce a new intelligent machine that can react in a similar way to human intelligence. Artificial intelligence is to study the design principles and implementation methods of various intelligent machines, so that machines have the functions of perception, reasoning and decision-making.

[0073] Artificial intelligence technology is a comprehensive discipline, involving a wide range of fields, both hardware and software technologies. Artificial intelligence basic technologies generally include technologies such as sensors, special artificial intelligence chips, cloud computing, distributed storage, big data processing technology, operation / interaction system, mechatronics, etc. Artificial intelligence software technology mainly includes computer vision technology, speech processing technology, natural language processing technology, and machine learning / deep learning, etc.

[0074] Machine learning (ML) is a multi-disciplinary subject, involving probability theory, statistics, approximation theory, convex analysis, algorithm complexity theory, etc. It is a subject that studies how computers simulate or implement human learning behavior to acquire new knowledge or skills, and reorganize existing knowledge sub-models to continuously improve their performance. Machine learning is the core of artificial intelligence and the fundamental approach to making computers intelligent, and its applications are widespread in various fields of artificial intelligence. Machine learning and deep learning usually include artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and teaching learning, etc.

[0075] Embedded Coding: Embedded coding is mathematically represented as a corresponding relationship, that is, the data on the X space is mapped to the Y space through a function F, where the function F is a single function, the mapping result is structure preservation, the single function means that the mapped data is uniquely corresponding to the pre-mapped data, and the structure preservation means that the size relationship of the pre-mapped data is the same as that of the post-mapped data, for example, there are data X1 and X2 before mapping, and Y1 corresponding to X1 and Y2 corresponding to X2 are obtained after mapping. If the data X1 before mapping is greater than X2, then the data Y1 after mapping is greater than Y2 accordingly. For words, it is to map words to another space, which is convenient for subsequent machine learning and processing.

[0076] Attention weight: can represent the importance of certain data in the training or prediction process, the importance represents the size of the influence of the input data on the output data. The data with high importance has a higher value of the corresponding attention weight, and the data with low importance has a lower value of the corresponding attention weight. In different scenarios, the importance of data is not the same, and the process of training attention weight of the model is also the process of determining the importance of data.

[0077] Immune repertoire: at any specified time, the sum of all functional diversity immune cells in the circulatory system of a certain individual.

[0078] Immune cells: commonly known as white blood cells, including innate lymphocytes, various phagocytes, and lymphocytes capable of recognizing antigens and producing specific immune responses.

[0079] T cells: full name T-lymphocyte, derived from bone marrow multipotent stem cells (in embryonic period, derived from yolk sac and liver). In the embryonic period and the early stage of human body, a part of multipotent stem cells or pre-T cells in bone marrow migrate to thymus, and mature under the induction of thymus hormone, become T cells with immune activity.

[0080] TCR: T cell receptor (TCR) is a characteristic marker on the surface of all T cells, and the function of TCR is to recognize antigens.

[0081] B cells: full name B-lymphocyte, derived from bone marrow multipotent stem cells. The progenitor cells of B-lymphocytes exist in the hematopoietic cell islands of fetal liver (embryonic mice 14 days or 8-9 weeks of postnatal children), and the production and differentiation sites of B-lymphocytes are gradually replaced by bone marrow. Mature B cells mainly inhabit the lymph node cortex and the lymph node of the red pulp and white pulp of the spleen. B cells can differentiate into plasma cells under the stimulation of antigens, and plasma cells can synthesize and secrete antibodies (immunoglobulins), mainly executing the humoral immunity of the body.

[0082] BCR: B-cell receptor (BCR) is a molecule located on the surface of B cells responsible for specific recognition and binding of antigens, which is essentially a membrane surface immunoglobulin. BCR has antigen binding specificity.

[0083] Antigen: refers to all substances that can stimulate the body to produce specific immune response (humoral immunity and cellular immunity).

[0084] Cloud Technology refers to a kind of hosting technology that unifies a series of resources such as hardware, software and network in a wide area network or local area network to realize data calculation, storage, processing and sharing.

[0085] The technical scheme provided by the embodiments of the present application can also be combined with cloud technology, for example, deploying the state prediction model obtained by training in a cloud server. Among them, the medical cloud in the cloud technology refers to using "cloud computing" to create a medical health service cloud platform based on cloud computing, mobile technology, multimedia, 4G communication, big data and other new technologies, combined with medical technology, which realizes the sharing of medical resources and the expansion of medical scope.

[0086] It should be noted that the information (including but not limited to user device information, user personal information, etc.), data (including but not limited to data for analysis, stored data, displayed data, etc.) and signals involved in the present application are all authorized by the user or fully authorized by all parties, and the collection, use and processing of related data need to comply with relevant laws, regulations and standards of relevant countries and regions. For example, the genetic information involved in the present application is obtained under full authorization.

[0087] Figure 1 is a schematic diagram of an implementation environment of a state prediction method provided by the embodiments of the present application, see Figure 1 The implementation environment can include a terminal 110 and a server 140.

[0088] The terminal 110 is connected to the server 140 through a wireless network or a wired network. Optionally, the terminal 110 is a smart phone, a tablet computer, a notebook computer, a desktop computer, a smart watch, etc., but is not limited thereto. The terminal 110 is installed and runs an application program supporting state prediction.

[0089] The server 140 is a stand-alone physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms.

[0090] Optionally, the terminal 110 generally refers to one of multiple terminals, and the embodiments of the present application are only exemplified by the terminal 110.

[0091] Those skilled in the art can know that the number of the above terminals can be more or less. For example, the above terminals are only one, or the above terminals are dozens or hundreds, or more, and at this time, the above implementation environment also includes other terminals. The number of terminals and the type of equipment are not limited in the embodiments of the present application.

[0092] After introducing the implementation environment of the embodiments of the present application, the technical solutions provided by the embodiments of the present application will be described in combination with the above implementation environment. In the following description, the terminal is the terminal 110 in the above implementation environment, and the server is the server 140 in the above implementation environment.

[0093] The state prediction method provided by the embodiments of the present application can be applied in the fields of scientific research and disease prediction, such as in the scene of determining the state (disease) of a target object (person). Through the technical solutions provided by the embodiments of the present application, the technical personnel upload the gene information, sequence information and three-dimensional structure features of multiple immune cell receptors in the immune repertoire of the target object to the server through the terminal. The server processes the gene information, sequence information and three-dimensional structure features of the multiple immune cell receptors through the trained state prediction model to obtain the receptor features of each immune cell receptor, wherein the gene information of each immune cell receptor includes the VDJ information of the immune cell receptor, the sequence information is the amino acid sequence of each immune cell receptor, and the three-dimensional structure features are used to represent the three-dimensional structure of each immune cell receptor. The server performs state prediction based on the receptor features of each immune cell receptor through the state prediction model, and outputs the state corresponding to each immune cell receptor, which is the prediction of the receptor level. The server determines at least one target state of the target object based on the states corresponding to the multiple immune cell receptors through the state prediction model, and the target state is the prediction of the individual level. In the case that the target object is a person, the at least one target state is at least one disease that the person may have, and the doctor can formulate a corresponding treatment plan based on the output of the state prediction model.

[0094] After introducing the implementation environment and application scenario of the embodiments of the present application, the state prediction method provided by the embodiments of the present application is described below. The technical solutions provided by the embodiments of the present application can be executed by a terminal or a server, or executed by a terminal and a server together. In the following description, the execution subject is taken as the server for example, and the description is made with reference to Figure 2 The method comprises the following steps.

[0095] 201. The server inputs the gene information, sequence information and three-dimensional structure features of a plurality of immune cell receptors in the immune repertoire of the target object into the state prediction model.

[0096] The target object is an object to be subjected to state prediction, the immune repertoire is the sum of immune cells in the circulatory system of the target object, and the immune cell receptor is a T cell receptor or a B cell receptor. In some embodiments, the gene information of the immune cell receptor comprises VDJ information of the immune cell receptor, wherein V represents a coding variable region, D represents a coding hypervariable region, and J represents a coding crosslinking region. The sequence information of the immune cell receptor is the amino acid sequence of the immune cell receptor. The three-dimensional structure features of the immune cell receptor are determined based on the three-dimensional structure of the immune cell receptor, wherein the three-dimensional structure is used to represent the positions of a plurality of amino acids in the immune cell receptor, and the three-dimensional structure features can reflect the three-dimensional structure of the immune cell receptor as a whole. The state prediction model is a model trained based on the gene information, sequence information and three-dimensional structure features of a plurality of immune cell receptors of a sample object, and has the function of corresponding to the state of an object.

[0097] 202. The server extracts features from the gene information and sequence information of the plurality of immune cell receptors by the state prediction model to obtain gene features and sequence features of each immune cell receptor.

[0098] The process of extracting features from the gene information and sequence information of each immune cell receptor is also the process of abstractly expressing the gene information and sequence information of each immune cell receptor, and the obtained gene features and sequence features can represent the gene information and sequence information of each immune cell receptor and facilitate subsequent processing by the server.

[0099] 203. The server fuses the gene features, sequence features and three-dimensional structure features of each immune cell receptor by the state prediction model to obtain receptor features of each immune cell receptor.

[0100] The receptor features of each immune cell receptor are obtained by fusing the gene features, sequence features and three-dimensional structure features, and can represent the immune cell receptor from the aspects of gene, sequence and structure, so the expression ability of the receptor features is strong.

[0101] 204、The server performs state prediction based on the receptor characteristics of the immune cell receptors through the state prediction model, and outputs at least one target state of the target object.

[0102] The target state is a state that the target object may be in, for example, in the case of the target object being a human, the target state is a disease that the human may have.

[0103] The state prediction model provided by the embodiments of the present application extracts the gene information and sequence of the immune cell receptors to obtain the gene characteristics and sequence characteristics of the immune cell receptors. In the process of obtaining the receptor characteristics of the immune cell receptors, the gene characteristics, sequence characteristics and three-dimensional structure characteristics are fused. The introduction of the three-dimensional structure characteristics enriches the content of the receptor characteristics and improves the expression ability of the receptor characteristics, so that the accuracy of the state prediction of the target object based on the receptor characteristics is higher.

[0104] The above steps 201-204 are a simple description of the state prediction method provided by the embodiments of the present application. The state prediction method provided by the embodiments of the present application will be further described below with reference to some examples, see Figure 3 Taking the server as an example, the method comprises the following steps.

[0105] 301、The server obtains the three-dimensional structure characteristics of a plurality of immune cell receptors in the immune repertoire of the target object.

[0106] The immune cell receptor is a T cell receptor or a B cell receptor, and the immune cell receptor is used to recognize and specifically bind to an antigen, thereby activating the immune system. The immune cell receptor is a protein, and the protein includes a plurality of amino acids. The three-dimensional structure characteristics of the immune cell receptor are used to represent the positions of the plurality of amino acids of the immune cell receptor in space.

[0107] In one possible implementation, for any immune cell receptor in the plurality of immune cell receptors, the server obtains a target amino acid sequence of the immune cell receptor, the target amino acid sequence including a CDR3 region of the immune cell receptor. The server performs multiple sequence alignment on the target amino acid sequence of the immune cell receptor to obtain at least one reference amino acid sequence, the similarity between the target amino acid sequence and the reference amino acid sequence meeting a similarity condition. The server obtains a homologous template corresponding to the target amino acid sequence, the homologous template including structure information of a homologous sequence of the target amino acid sequence. The server performs multiple iterations based on the target amino acid sequence, the at least one reference amino acid sequence and the homologous template to obtain the three-dimensional structure characteristics of the immune cell receptor.

[0108] The immune cell receptor has a complementary determining region (CDR) including three sub-regions CDR1, CDR2 and CDR3, wherein CDR3 is the most variable and plays a key role in antigen recognition.

[0109] In this implementation, the server can determine the three-dimensional structure characteristics of each immune cell receptor based on the target amino acid sequence of the immune cell receptor, without the need to observe through other devices such as cryo-EM, thereby improving the acquisition efficiency of the three-dimensional structure characteristics and reducing the acquisition cost of the three-dimensional structure characteristics.

[0110] For example, for any of the plurality of immune cell receptors, the server obtains sequencing data of the immune cell receptor, the sequencing data comprising a plurality of amino acids of the immune cell receptor and an arrangement order of the plurality of amino acids, the sequencing data being obtained by a gene sequencing device, which is not limited by the embodiments of the present application. The server preprocesses the sequencing data of the immune cell receptor to obtain reference sequencing data of the immune cell receptor, wherein the preprocessing of the sequencing data comprises eliminating error data in the sequencing data and converting the sequencing data into a format convenient for the server to process, etc., and the rules of preprocessing are set by the technicians according to the actual situation, which is not limited by the embodiments of the present application. The server performs quality control on the reference sequencing data to obtain target sequencing data of the immune cell receptor, wherein the quality control of the reference sequencing data comprises filtering out dead cells, background estimation, paired chains, dextramer signal correction, Log-rank test and receptor gene aggregation. The server extracts an amino acid sequence of a target length containing a CDR3 region from the target sequencing data, the amino acid sequence of the target length containing the CDR3 region being also a target amino acid sequence, wherein the target length is set by the technicians according to the actual situation, such as being greater than 50 amino acids, etc., which is not limited by the embodiments of the present application. The server searches in a gene database based on the target amino acid sequence to obtain at least one reference amino acid sequence, the at least one reference amino acid sequence being also an amino acid sequence having a similarity greater than or equal to a similarity threshold with the target amino acid sequence, the similarity between amino acid sequences being determined by comparing the types and arrangement orders of amino acids in the amino acid sequences, and the multiple sequence alignment is also called multiple sequence alignment, which is used to extract and input similar sequences of amino acid sequences from a large database, and align them incidentally. Since the amino acid sequences with similar sequences generally have similar folding modes, multiple sequence alignment can add similar sequence structure information in the characteristics. The server searches in a structure database based on the target amino acid sequence to obtain a homologous template corresponding to the target amino acid sequence, the homologous template comprising structure information of a homologous sequence of the target amino acid sequence. The server performs multiple rounds of iterative encoding on the target amino acid sequence, the at least one reference amino acid sequence and the homologous template based on an attention mechanism to obtain distance distributions between each pair of amino acids in the target amino acid sequence and angles of chemical bonds connecting them.The server encodes the distance distribution between each pair of amino acids in the target amino acid sequence of the immune cell receptor and the angle of the chemical bond connecting them using an attention mechanism, and outputs the three-dimensional structure information of the immune cell receptor, wherein the three-dimensional structure information of the immune cell receptor includes the three-dimensional positions of the plurality of amino acids in the immune cell receptor. The server extracts features from the three-dimensional structure of the immune cell receptor, such as processing the immune cell receptor using a graph network to obtain three-dimensional structure features of the immune cell receptor.

[0111] In order to more clearly illustrate the above-mentioned embodiments, the following will be combined with Figure 4 The above-mentioned embodiments are described.

[0112] Referring to Figure 4 For any immune cell receptor in the plurality of immune cell receptors, the server pre-processes 401 the sequencing data of the immune cell receptor to obtain reference sequencing data of the immune cell receptor. The server performs quality control 402 on the reference sequencing data to obtain target sequencing data of the immune cell receptor, wherein the quality control 402 includes dead cell removal 4021, background estimation 4022, chain pairing 4023, signal correction 4024, Log-rank test 4025 and receptor gene aggregation 4026. The server performs sequence clipping 403 on the target sequencing data to obtain a target amino acid sequence. The server performs multiple sequence alignment 404 based on the target amino acid sequence to obtain at least one reference amino acid sequence. The server searches in a structure database based on the target amino acid sequence to obtain a homologous template corresponding to the target amino acid sequence. The server encodes 405 the target amino acid sequence, the at least one reference amino acid sequence and the homologous template through a state based on an attention mechanism to obtain three-dimensional structure information of the immune cell receptor.

[0113] The above-mentioned embodiments are methods for the server to determine the three-dimensional structure features of the immune cell receptor based on the target amino acid sequence of the immune cell receptor. In other possible embodiments, the server can use a trained structure prediction model to obtain three-dimensional structure features based on the amino acid sequence, wherein the structure prediction model includes RoseTTAFold, AlphaFold, AlphaFold2 and other models. Of course, as science and technology develop, other structure prediction models can also be used, and the present application embodiments are not limited thereto.

[0114] The following describes a method for the server to obtain three-dimensional structure features of the immune cell receptor based on three-dimensional structure information of the immune cell receptor, wherein the three-dimensional structure information includes three-dimensional positions of a plurality of amino acids in the immune cell receptor.

[0115] In a possible implementation, for any immune cell receptor in the plurality of immune cell receptors, the server obtains three-dimensional structure information of the immune cell receptor, the three-dimensional structure information comprising three-dimensional coordinates of a plurality of amino acids in the immune cell receptor. The server performs graph convolution on the three-dimensional structure information of the immune cell receptor to obtain a three-dimensional structure feature of the immune cell receptor.

[0116] In some embodiments, the three-dimensional structure information is obtained from images taken by cryo-electron microscopy, or is obtained from a structure prediction model based on an amino acid sequence of the immune cell receptor, without any limitation in the embodiments of this application. Graph convolution is the full name of a graph convolutional network (GCN), which is used to extract features of a graph. In the embodiments of this application, the nodes in the graph are the amino acids in the immune cell receptor, and the connections in the graph are used to represent the relative positional relationship between the amino acids.

[0117] In this implementation, the server directly performs graph convolution on the three-dimensional structure information of the immune cell receptor to obtain the three-dimensional structure feature of the immune cell receptor, without the need to first determine the three-dimensional structure information of the immune cell receptor, and the efficiency of determining the three-dimensional structure feature is relatively high.

[0118] For example, for any immune cell receptor in the plurality of immune cell receptors, the server obtains three-dimensional structure information of the immune cell receptor. The server generates a three-dimensional structure graph of the immune cell receptor based on the three-dimensional structure information, the nodes in the three-dimensional structure graph corresponding to the amino acids in the immune cell receptor, the connections in the three-dimensional structure graph being used to represent the connection relationship between the amino acids, and the node features of the nodes in the three-dimensional structure graph comprising the types and three-dimensional coordinates of the corresponding amino acids. The server performs graph convolution on the three-dimensional structure graph to obtain a three-dimensional structure feature of the immune cell receptor.

[0119] In a possible implementation, for any immune cell receptor in the plurality of immune cell receptors, the server obtains three-dimensional structure information of the immune cell receptor, the three-dimensional structure information comprising three-dimensional coordinates of a plurality of amino acids in the immune cell receptor. The server encodes the three-dimensional structure information of the immune cell receptor based on an attention mechanism to obtain a three-dimensional structure feature of the immune cell receptor.

[0120] In this implementation, the server directly encodes the three-dimensional structure information of the immune cell receptor based on the attention mechanism to obtain the three-dimensional structure feature of the immune cell receptor, without the need to first determine the three-dimensional structure information of the immune cell receptor, and the efficiency of determining the three-dimensional structure feature is relatively high.

[0121] For example, the server obtains three-dimensional structure information of any one of the plurality of immune cell receptors. The server encodes a plurality of amino acids in the three-dimensional structure information by embedding to obtain a plurality of amino acid embedding features. The process of encoding a plurality of amino acids by embedding is to represent a plurality of amino acids in a discrete form, which facilitates subsequent processing of the server. The server encodes the plurality of amino acid embedding features based on the three-dimensional structure information by using an attention mechanism to obtain attention weights of the plurality of amino acids. The server fuses the plurality of amino acid embedding features based on the attention weights of the plurality of amino acids to obtain three-dimensional structure features of the immune cell receptor. In some embodiments, the server can use an encoder of a Transformer model to encode the three-dimensional structure information of the immune cell receptor to obtain the three-dimensional structure features of the immune cell receptor.

[0122] It should be noted that the above two embodiments are described by taking an example of the server encoding the three-dimensional structure information of the immune cell receptor by using graph convolution and attention mechanism to obtain three-dimensional structure features. In other possible embodiments, the server can also encode the three-dimensional structure information of the immune cell receptor by using other models, which are not limited in the embodiments of the present application.

[0123] It should be noted that the above several embodiments are described by taking an example of the server processing one immune cell receptor to obtain three-dimensional structure features of the immune cell receptor. For other immune cell receptors in the plurality of immune cell receptors, the server obtains three-dimensional structure features in the same inventive concept as the above several embodiments, and the implementation process is not described again.

[0124] It should be noted that the above step 301 is an optional step.

[0125] 302、The server inputs the gene information, sequence information, and three-dimensional structure features of a plurality of immune cell receptors in the immune repertoire of the target object into a state prediction model.

[0126] The gene information of the immune cell receptor includes VDJ information of the immune cell receptor, where V represents a coding variable region, D represents a coding variable region, and J represents a coding cross-linking region. The sequence information of the immune cell receptor is an amino acid sequence of the immune cell receptor, for example, AEGAL is an amino acid sequence, where A represents alanine (Alanine), E represents glutamic acid (Glutamic acid), G represents glycine (Glycine), and L represents leucine (Leucine). The immune cell receptor is a kind of protein, and the amino acid sequence is also called one-dimensional structure of the protein.

[0127] In a possible implementation, the state prediction model comprises three information encoding channels, wherein the first information encoding channel is a gene information encoding channel, the gene information encoding channel comprises a gene encoder configured to encode gene information; the second information encoding channel is a sequence information encoding channel, the sequence information encoding channel comprises a sequence encoder configured to encode sequence information; and the third information encoding channel is a structure feature encoding channel, the structure feature encoding channel comprises a structure encoder configured to encode structure features. The server inputs the gene information of the immune cell receptor into the gene information encoding channel of the antigen prediction model, and subsequently encodes the gene information by the gene encoder in the gene information encoding channel. The server inputs the sequence information of the immune cell receptor into the sequence information encoding channel of the antigen prediction model, and subsequently encodes the sequence information by the sequence encoder in the sequence information encoding channel. The server inputs the three-dimensional structure features of the immune cell receptor into the structure feature encoding channel, and subsequently encodes the three-dimensional structure features by the structure encoder in the structure feature encoding channel.

[0128] In some embodiments, for any immune cell receptor in the plurality of immune cell receptors, before inputting the sequence information of the immune cell receptor into the state prediction model, the server can also pre-process the sequence information of the immune cell receptor to ensure that the lengths of the sequence information input into the state prediction model are all the same. In the case that the length of the sequence information of the immune cell receptor is greater than a length threshold, the server truncates the part of the sequence information of the immune cell receptor with a length greater than or equal to the length threshold to obtain sequence information with a length of the length threshold, and subsequently inputs the truncated sequence information into the state prediction model. In the case that the length of the sequence information of the immune cell receptor is less than the length threshold, the server pads the sequence information of the immune cell receptor with a target symbol to obtain sequence information with a length of the length threshold, and subsequently inputs the truncated sequence information into the state prediction model, wherein the target symbol is set by the technician according to the actual situation, such as 0.

[0129] It should be noted that the above steps 301-302 are described by taking the server as an example of pre-acquiring the three-dimensional structure features of the plurality of immune cell receptors. In other possible implementations, the server can also pre-acquire the three-dimensional structure information of the plurality of immune cell receptors, input the three-dimensional structure information of the plurality of immune cell receptors into the structure feature encoding channel of the state prediction model, and subsequently acquire the three-dimensional structure features of the plurality of immune cell receptors by the structure encoder of the structure feature encoding channel, which is not limited in the embodiments of the present application.

[0130] In addition, steps 301-302 are described by taking an example of the server obtaining three-dimensional structural features of a plurality of immune cell receptors, and inputting gene information, sequence information, and three-dimensional structural features of each immune cell receptor into a state prediction model. In other possible implementations, in the case where the server does not obtain three-dimensional structural features of any immune cell receptor, the gene information and sequence information of the immune cell receptor can also be input into the state prediction model.

[0131] 303. The server extracts features of the gene information and the sequence information of the plurality of immune cell receptors by the state prediction model to obtain gene features and sequence features of each immune cell receptor.

[0132] The process of extracting features of the gene information and the sequence information of each immune cell receptor is a process of abstractly representing the gene information and the sequence information of each immune cell receptor, and the obtained gene features and sequence features can represent the gene information and the sequence information of each immune cell receptor and facilitate subsequent processing by the server.

[0133] In one possible implementation, the state prediction model includes a gene encoder and a sequence encoder. For any immune cell receptor in the plurality of immune cell receptors, the server encodes VDJ information of the immune cell receptor by the gene encoder of the state prediction model to obtain gene features of the immune cell receptor, where V is a variable region, D is a hypervariable region, and J is a cross-linking region. The server encodes an amino acid sequence of the immune cell receptor by the sequence encoder of the state prediction model to obtain sequence features of the immune cell receptor.

[0134] In this implementation, the server can encode the gene information and the sequence information of the immune cell receptor by the gene encoder and the sequence encoder of the state prediction model, i.e., extract features of the gene information and the sequence information, and the obtained gene features and sequence features can represent the immune cell receptor from different dimensions.

[0135] To more clearly illustrate the above implementation, the above implementation will be described in two parts. It should be noted that in the following description, an example of the server processing any immune cell receptor in the plurality of immune cell receptors is taken to illustrate the processing method for other immune cell receptors in the plurality of immune cell receptors, which belongs to the same inventive concept.

[0136] In the first part, the server encodes VDJ information of the immune cell receptor by the gene encoder of the state prediction model to obtain gene features of the immune cell receptor.

[0137] In a possible implementation, in the case that the immune cell receptor is a B cell receptor, the server encodes VJ information of a light chain and VDJ information of a heavy chain of the immune cell receptor by a gene encoder of the state prediction model to obtain a gene feature of the immune cell receptor.

[0138] The B cell receptor includes two identical heavy chains (H chain) and two identical light chains (L chain), and the two heavy chains and the two light chains are connected into a four-chain structure through inter-chain disulfide bonds. The molecular weight of the heavy chain is about 50-75 kD, which is composed of 450-550 amino acid residues. The molecular weight of the light chain is about 25 kD, which is composed of 214 amino acid residues.

[0139] In order to more clearly illustrate the above implementation, the above implementation will be illustrated by three examples.

[0140] In example 1, the server fully connects VJ information of a light chain and VDJ information of a heavy chain of the immune cell receptor by a gene encoder of the state prediction model to obtain a gene feature of the immune cell receptor, and the gene feature of the immune cell receptor includes a light chain gene feature of the immune cell receptor and a heavy chain gene feature of the immune cell receptor.

[0141] In a possible implementation, the state prediction model includes two gene encoders, the server splices VJ information of a light chain of the B cell receptor by a first gene encoder of the state prediction model to obtain light chain gene information of the B cell receptor. The server splices VDJ information of the light chain of the B cell receptor by a second gene encoder of the state prediction model to obtain heavy chain gene information of the B cell receptor. The server fully connects the light chain gene information of the B cell receptor twice by the first gene encoder of the state prediction model to obtain a light chain gene feature of the B cell receptor. The server fully connects the heavy chain gene information of the B cell receptor twice by the second gene encoder of the state prediction model to obtain a heavy chain gene feature of the B cell receptor. The light chain gene feature and the heavy chain gene feature of the B cell receptor constitute a gene feature of the B cell receptor.

[0142] In example 2, the server convolves VJ information of a light chain and VDJ information of a heavy chain of the immune cell receptor by a gene encoder of the state prediction model to obtain a gene feature of the immune cell receptor, and the gene feature of the immune cell receptor includes a light chain gene feature of the immune cell receptor and a heavy chain gene feature of the immune cell receptor.

[0143] In a possible implementation, the state prediction model includes two gene encoders. The server splices the VJ information of the light chain of the B cell receptor to obtain light chain gene information of the B cell receptor through a first gene encoder of the state prediction model. The server splices the VDJ information of the light chain of the B cell receptor to obtain heavy chain gene information of the B cell receptor through a second gene encoder of the state prediction model. The server performs twice convolution on the light chain gene information of the B cell receptor through the first gene encoder of the state prediction model to obtain light chain gene features of the B cell receptor. The server performs twice convolution on the heavy chain gene information of the B cell receptor through the second gene encoder of the state prediction model to obtain heavy chain gene features of the B cell receptor. The light chain gene features and the heavy chain gene features of the B cell receptor constitute gene features of the B cell receptor.

[0144] In example 3, the server encodes the VJ information of the light chain and the VDJ information of the heavy chain of the immune cell receptor based on an attention mechanism through a gene encoder of the state prediction model to obtain gene features of the immune cell receptor, where the gene features of the immune cell receptor include light chain gene features of the immune cell receptor and heavy chain gene features of the immune cell receptor.

[0145] In a possible implementation, the state prediction model includes two gene encoders. The server splices the VJ information of the light chain of the B cell receptor to obtain light chain gene information of the B cell receptor through a first gene encoder of the state prediction model. The server splices the VDJ information of the light chain of the B cell receptor to obtain heavy chain gene information of the B cell receptor through a second gene encoder of the state prediction model. The server encodes the light chain gene information of the B cell receptor based on an attention mechanism through the first gene encoder of the state prediction model to obtain light chain gene features of the B cell receptor. The server encodes the heavy chain gene information of the B cell receptor based on an attention mechanism through the second gene encoder of the state prediction model to obtain heavy chain gene features of the B cell receptor. The light chain gene features and the heavy chain gene features of the B cell receptor constitute gene features of the B cell receptor.

[0146] The above is described by taking the immune cell receptor as the B cell receptor. The following is described by taking the immune cell receptor as the T cell receptor.

[0147] In a possible implementation, in the case where the immune cell receptor is the T cell receptor, the server encodes the VJ information of the α chain and the VDJ information of the β chain of the immune cell receptor through a gene encoder of the state prediction model to obtain gene features of the immune cell receptor.

[0148] Some of the T cell receptors include an a chain and a β chain, and such T cell receptors are also referred to as aβ-TCR. Some other T cell receptors include a γ chain and a δ chain, and such T cell receptors are also referred to as γδ-TCR. Since the number of aβ-TCR is much larger than the number of γδ-TCR in the human body, the following description is given by taking the T cell receptor as an aβ-TCR. For a γδ-TCR, the structure is similar to that of an aβ-TCR, and the processing mode belongs to the same inventive concept, and the implementation process is described below.

[0149] To make the above-mentioned embodiments clearer, the following describes the above-mentioned embodiments by taking three examples.

[0150] In Example 1, a server performs full connection on VJ information of an a chain and VDJ information of a β chain of an immune cell receptor by using a gene encoder of a state prediction model to obtain gene features of the immune cell receptor, and the gene features of the immune cell receptor include a chain gene features of the immune cell receptor and β chain gene features of the immune cell receptor.

[0151] In a possible implementation, the state prediction model includes two gene encoders. The server splices VJ information of an a chain of a T cell receptor by using a first gene encoder of the state prediction model to obtain a chain gene information of the T cell receptor. The server splices VDJ information of the a chain of the T cell receptor by using a second gene encoder of the state prediction model to obtain β chain gene information of the T cell receptor. The server performs twice full connection on the a chain gene information of the T cell receptor by using the first gene encoder of the state prediction model to obtain a chain gene features of the T cell receptor. The server performs twice full connection on the β chain gene information of the T cell receptor by using the second gene encoder of the state prediction model to obtain β chain gene features of the T cell receptor. The a chain gene features and the β chain gene features of the T cell receptor constitute gene features of the T cell receptor.

[0152] In Example 2, a server performs convolution on VJ information of an a chain and VDJ information of a β chain of an immune cell receptor by using a gene encoder of a state prediction model to obtain gene features of the immune cell receptor, and the gene features of the immune cell receptor include a chain gene features of the immune cell receptor and β chain gene features of the immune cell receptor.

[0153] In a possible implementation, the state prediction model comprises two gene encoders, the server splices the VJ information of the alpha chain of the T cell receptor by a first gene encoder of the state prediction model to obtain alpha chain gene information of the T cell receptor. The server splices the VDJ information of the alpha chain of the T cell receptor by a second gene encoder of the state prediction model to obtain beta chain gene information of the T cell receptor. The server performs twice convolution on the alpha chain gene information of the T cell receptor by the first gene encoder of the state prediction model to obtain alpha chain gene features of the T cell receptor. The server performs twice convolution on the beta chain gene information of the T cell receptor by the second gene encoder of the state prediction model to obtain beta chain gene features of the T cell receptor. The alpha chain gene features and the beta chain gene features of the T cell receptor constitute gene features of the T cell receptor.

[0154] In example 3, the server encodes the VJ information of the alpha chain and the VDJ information of the beta chain of the immune cell receptor by a gene encoder of the state prediction model based on an attention mechanism to obtain gene features of the immune cell receptor, the gene features of the immune cell receptor comprising alpha chain gene features of the immune cell receptor and beta chain gene features of the immune cell receptor.

[0155] In a possible implementation, the state prediction model comprises two gene encoders, the server splices the VJ information of the alpha chain of the T cell receptor by a first gene encoder of the state prediction model to obtain alpha chain gene information of the T cell receptor. The server splices the VDJ information of the alpha chain of the T cell receptor by a second gene encoder of the state prediction model to obtain beta chain gene information of the T cell receptor. The server encodes the alpha chain gene information of the T cell receptor by the first gene encoder of the state prediction model based on an attention mechanism to obtain alpha chain gene features of the T cell receptor. The server encodes the beta chain gene information of the T cell receptor by the second gene encoder of the state prediction model based on an attention mechanism to obtain beta chain gene features of the T cell receptor. The alpha chain gene features and the beta chain gene features of the T cell receptor constitute gene features of the T cell receptor.

[0156] In the second part, the server encodes the amino acid sequence of the immune cell receptor by a sequence encoder of the state prediction model to obtain sequence features of the immune cell receptor.

[0157] In a possible implementation, in a case where the immune cell receptor is a B cell receptor, the server encodes, by a sequence encoder of the state prediction model, amino acid sequences of a light chain and a heavy chain of the B cell receptor based on an attention mechanism to obtain sequence features of the B cell receptor, the sequence features of the B cell receptor including light chain sequence features and heavy chain sequence features of the B cell receptor. In some embodiments, the sequence encoder is an encoder of a Transformer model.

[0158] For example, the state prediction model includes two sequence encoders, in a case where the immune cell receptor is a B cell receptor, the server embeds, by a first sequence encoder of the state prediction model, amino acid sequences of a light chain of the B cell receptor to obtain light chain embedding features of the B cell receptor, one light chain embedding feature corresponding to one amino acid on the light chain. The server encodes, by the first sequence encoder, the light chain embedding features based on orders of the amino acids in the amino acid sequences of the B cell receptor to obtain attention weights corresponding to the light chain embedding features. The server fuses, by the first sequence encoder, the light chain embedding features based on the attention weights corresponding to the light chain embedding features to obtain the light chain sequence features of the B cell receptor. The server embeds, by a second sequence encoder of the state prediction model, amino acid sequences of a heavy chain of the B cell receptor to obtain heavy chain embedding features of the B cell receptor, one heavy chain embedding feature corresponding to one amino acid on the heavy chain. The server encodes, by the second sequence encoder, the heavy chain embedding features based on orders of the amino acids in the amino acid sequences of the B cell receptor to obtain attention weights corresponding to the heavy chain embedding features. The server fuses, by the second sequence encoder, the heavy chain embedding features based on the attention weights corresponding to the heavy chain embedding features to obtain the heavy chain sequence features of the B cell receptor. The light chain sequence features of the B cell receptor and the heavy chain sequence features of the B cell receptor constitute the sequence features of the B cell receptor. In some embodiments, the embedding encoding can be in a one-hot manner or other manners, which are not limited in the embodiments of the present application.

[0159] In a possible implementation, in a case where the immune cell receptor is a T cell receptor, the server encodes, by a sequence encoder of the state prediction model, amino acid sequences of an alpha chain and a beta chain of the T cell receptor based on an attention mechanism to obtain sequence features of the T cell receptor, the sequence features of the T cell receptor including alpha chain sequence features and beta chain sequence features of the T cell receptor.

[0160] For example, the state prediction model includes two sequence encoders. When the immune cell receptor is a T cell receptor, the server embeds the amino acid sequence of the a chain of the T cell receptor by a first sequence encoder of the state prediction model to obtain a chain embedding features, one a chain embedding feature corresponding to one amino acid on the a chain. The server encodes the a chain embedding features based on the order of the plurality of amino acids in the amino acid sequence of the T cell receptor by the first sequence encoder to obtain attention weights corresponding to each a chain embedding feature. The server fuses the a chain embedding features by weighting based on the attention weights corresponding to each a chain embedding feature by the first sequence encoder to obtain a chain sequence features of the T cell receptor. The server embeds the amino acid sequence of the b chain of the T cell receptor by a second sequence encoder of the state prediction model to obtain b chain embedding features, one b chain embedding feature corresponding to one amino acid on the b chain. The server encodes the b chain embedding features based on the order of the plurality of amino acids in the amino acid sequence of the T cell receptor by the second sequence encoder to obtain attention weights corresponding to each b chain embedding feature. The server fuses the b chain embedding features by weighting based on the attention weights corresponding to each b chain embedding feature by the second sequence encoder to obtain b chain sequence features of the T cell receptor. The light chain sequence features of the T cell receptor and the heavy chain sequence features of the T cell receptor constitute the sequence features of the T cell receptor.

[0161] 304、The server fuses the gene features, the sequence features, and the three-dimensional structure features of each immune cell receptor by the state prediction model to obtain receptor features of each immune cell receptor.

[0162] The receptor features of the immune cell receptor are obtained by fusing the gene features, the sequence features, and the three-dimensional structure features, and can represent the immune cell receptor from the aspects of gene, sequence, and structure. The receptor features can more completely represent the immune cell receptor.

[0163] In a possible implementation, for any immune cell receptor in the plurality of immune cell receptors, the server splices the gene feature and the sequence feature of the immune cell receptor by a feature fusion module of the state prediction model to obtain a gene sequence fusion feature of the immune cell receptor. The server fuses the gene sequence fusion feature and the three-dimensional structure feature of the immune cell receptor by weighting based on a gated attention mechanism by the feature fusion module of the state prediction model to obtain a receptor feature of the immune cell receptor. It should be noted that in the following description, the server is taken as an example to process any immune cell receptor in the plurality of immune cell receptors, and the processing manner for other immune cell receptors in the plurality of immune cell receptors belongs to the same inventive concept as the method of processing the immune cell receptor.

[0164] In this implementation, the server can first fuse the gene feature and the sequence feature of the immune cell receptor by the feature fusion module to obtain the gene sequence fusion feature of the immune cell receptor. The server then fuses the sequence fusion feature and the three-dimensional structure feature based on the gated attention mechanism to finally obtain the receptor feature of the immune cell receptor. The introduction of the gated attention mechanism enables the model to pay more attention to the content with higher importance. Through the feature fusion manner provided in the above implementation, the gene feature, the sequence feature and the three-dimensional structure feature can be organically combined, and the obtained receptor feature has stronger expression ability.

[0165] In the case where the immune cell receptor is a B cell receptor, the gene feature of the B cell receptor includes a light chain gene feature and a heavy chain gene feature of the B cell receptor, and the sequence feature of the B cell receptor includes a light chain sequence feature and a heavy chain sequence feature of the B cell receptor. The server adds the light chain gene feature of the B cell receptor and the light chain sequence feature of the B cell receptor by the feature fusion module to obtain a light chain gene sequence feature of the B cell receptor. The server adds the heavy chain gene feature of the B cell receptor and the heavy chain sequence feature of the B cell receptor by the feature fusion module to obtain a heavy chain gene sequence feature of the B cell receptor. The server splices the light chain gene sequence feature and the heavy chain gene sequence feature of the B cell receptor by the feature fusion module to obtain a gene sequence fusion feature of the B cell receptor. The server encodes the gene sequence fusion feature and the three-dimensional structure feature of the B cell receptor by the feature fusion module using an attention mechanism to obtain a first attention weight of the gene sequence fusion feature encoding the three-dimensional structure feature and a second attention weight of the three-dimensional structure feature encoding the gene sequence fusion feature. The server processes the first attention weight and the second attention weight by the feature fusion module using a gating function to obtain a first gating weight and a second gating weight, which are used to control the flow of information during feature fusion. The server weights and fuses the gene sequence fusion feature and the three-dimensional structure feature of the B cell receptor by the feature fusion module using the first gating weight to obtain a target gene sequence fusion feature of the B cell receptor. In some embodiments, the first gating weight is multiplied by the three-dimensional structure feature and then added to the gene sequence fusion feature to obtain the target gene sequence fusion feature. The server weights and fuses the gene sequence fusion feature and the three-dimensional structure feature of the B cell receptor by the feature fusion module using the second gating weight to obtain a target three-dimensional structure feature of the B cell receptor. In some embodiments, the second gating weight is multiplied by the gene sequence fusion feature and then added to the three-dimensional structure feature to obtain the target three-dimensional structure feature. The server tensor fuses the target gene sequence fusion feature and the target three-dimensional structure feature by the feature fusion module, such as multiplying the target gene sequence fusion feature and the target three-dimensional structure, to obtain an initial receptor feature of the B cell receptor. The server performs at least two full connections on the initial receptor feature of the B cell receptor by the feature fusion module to obtain a receptor feature of the B cell receptor.

[0166] In the case where the immune cell receptor is a T cell receptor, the gene feature of the T cell receptor includes an alpha chain gene feature and a beta chain gene feature of the T cell receptor, and the sequence feature of the T cell receptor includes an alpha chain sequence feature and a beta chain sequence feature of the T cell receptor. The server adds the alpha chain gene feature of the T cell receptor and the alpha chain sequence feature of the T cell receptor by the feature fusion module to obtain an alpha chain gene sequence feature of the T cell receptor. The server adds the beta chain gene feature of the T cell receptor and the beta chain sequence feature of the T cell receptor by the feature fusion module to obtain a beta chain gene sequence feature of the T cell receptor. The server splices the alpha chain gene sequence feature and the beta chain gene sequence feature of the T cell receptor by the feature fusion module to obtain a gene sequence fusion feature of the T cell receptor. The server encodes the gene sequence fusion feature and the three-dimensional structure feature of the T cell receptor by the feature fusion module using an attention mechanism to obtain a third attention weight of encoding the gene sequence fusion feature on the three-dimensional structure feature and a fourth attention weight of encoding the three-dimensional structure feature on the gene sequence fusion feature. The server processes the third attention weight and the fourth attention weight by the feature fusion module using a gating function to obtain a third gating weight and a fourth gating weight, which are used to control the flow of information during feature fusion. The server weights and fuses the gene sequence fusion feature and the three-dimensional structure feature of the T cell receptor by the feature fusion module using the third gating weight to obtain a target gene sequence fusion feature of the T cell receptor, which in some embodiments is obtained by multiplying the third gating weight by the three-dimensional structure feature and adding the result to the gene sequence fusion feature. The server weights and fuses the gene sequence fusion feature and the three-dimensional structure feature of the T cell receptor by the feature fusion module using the fourth gating weight to obtain a target three-dimensional structure feature of the T cell receptor, which in some embodiments is obtained by multiplying the fourth gating weight by the gene sequence fusion feature and adding the result to the three-dimensional structure feature. The server tensor fuses the target gene sequence fusion feature and the target three-dimensional structure feature by the feature fusion module, such as by multiplying the target gene sequence fusion feature by the target three-dimensional structure to obtain an initial receptor feature of the T cell receptor. The server performs at least two full connections on the initial receptor feature of the T cell receptor by the feature fusion module to obtain a receptor feature of the T cell receptor.

[0167] In a possible implementation, the server adds the gene feature and the sequence feature of the immune cell receptor by a feature fusion module of the state prediction model to obtain a gene sequence fusion feature of the immune cell receptor. The server splices and performs at least one full connection on the sequence feature and the three-dimensional structure feature of the immune cell receptor by the feature fusion module to obtain the receptor feature of the immune cell receptor.

[0168] In this implementation, the server can quickly fuse the gene feature, the sequence feature and the three-dimensional structure feature of the immune cell receptor by addition, splicing and full connection to obtain the receptor feature of the immune cell receptor by the feature fusion module, which is efficient.

[0169] In the case where the immune cell receptor is a B cell receptor, the gene feature of the B cell receptor includes a light chain gene feature and a heavy chain gene feature of the B cell receptor, and the sequence feature of the B cell receptor includes a light chain sequence feature and a heavy chain sequence feature of the B cell receptor. The server adds the light chain gene feature of the B cell receptor and the light chain sequence feature of the B cell receptor by the feature fusion module to obtain a light chain gene sequence feature of the B cell receptor. The server adds the heavy chain gene feature of the B cell receptor and the heavy chain sequence feature of the B cell receptor by the feature fusion module to obtain a heavy chain gene sequence feature of the B cell receptor. The light chain gene sequence feature and the heavy chain gene sequence feature of the B cell receptor constitute a gene sequence fusion feature of the B cell receptor. The server splices the gene sequence fusion feature and the three-dimensional structure feature of the B cell receptor by the feature fusion module to obtain an initial receptor feature of the B cell receptor. The server performs at least one full connection on the initial receptor feature of the B cell receptor by the feature fusion module to obtain the receptor feature of the B cell receptor.

[0170] In the case where the immune cell receptor is a T cell receptor, the genetic characteristics of the T cell receptor include the alpha chain genetic characteristics and the beta chain genetic characteristics of the T cell receptor, and the sequence characteristics of the T cell receptor include the alpha chain sequence characteristics and the beta chain sequence characteristics of the T cell receptor. The server adds the alpha chain genetic characteristics of the T cell receptor and the alpha chain sequence characteristics of the T cell receptor by the feature fusion module to obtain the alpha chain genetic sequence characteristics of the T cell receptor. The server adds the beta chain genetic characteristics of the T cell receptor and the beta chain sequence characteristics of the T cell receptor by the feature fusion module to obtain the beta chain genetic sequence characteristics of the T cell receptor. The alpha chain genetic sequence characteristics and the beta chain genetic sequence characteristics of the T cell receptor constitute the genetic sequence fusion characteristics of the T cell receptor. The server splices the genetic sequence fusion characteristics of the T cell receptor and the three-dimensional structure characteristics by the feature fusion module to obtain the initial receptor characteristics of the T cell receptor. The server performs at least one full connection on the initial receptor characteristics of the T cell receptor by the feature fusion module to obtain the receptor characteristics of the T cell receptor.

[0171] It should be noted that the above is an example of fusing the genetic characteristics, sequence characteristics and three-dimensional structure characteristics of the immune cell receptor by the server to obtain the receptor characteristics of the immune cell receptor. In other possible embodiments, in addition to fusing the genetic characteristics, sequence characteristics and three-dimensional structure characteristics of the immune cell receptor, the server can also fuse other information to obtain the receptor characteristics of the immune cell receptor, see the following embodiments.

[0172] In one possible embodiment, the server fuses the genetic characteristics, sequence characteristics, three-dimensional structure characteristics of each immune cell receptor and the physicochemical information of the amino acids in each immune cell receptor by the state prediction model to obtain the receptor characteristics of each immune cell receptor.

[0173] The physicochemical information of the amino acids in the immune cell receptor includes the physical characteristics and chemical characteristics of the amino acids, wherein the physical characteristics include basic composition and structure, solubility, melting point, boiling point, optical behavior and optical activity, etc. The chemical characteristics include acidity and basicity and hydrophobicity, etc. The introduction of the physicochemical information of the amino acids in the receptor characteristics of the immune cell receptor can improve the expression ability of the receptor characteristics, so that the receptor characteristics can more completely represent the immune cell receptor.

[0174] For example, for any immune cell receptor in the plurality of immune cell receptors, the server splices the genetic feature and the sequence feature of the immune cell receptor through the feature fusion module to obtain a gene sequence fusion feature of the immune cell receptor. The server fuses the gene sequence fusion feature and the three-dimensional structure feature of the immune cell receptor through the feature fusion module of the state prediction model based on a gated attention mechanism to obtain an initial receptor feature of the immune cell receptor. The server adds the initial receptor feature of the immune cell receptor and the physicochemical information of the amino acid in the immune cell receptor through the feature fusion module to obtain a receptor feature of the immune cell receptor.

[0175] 305、The server outputs the probability of each immune cell receptor corresponding to a plurality of candidate states through the classification module of the state prediction model.

[0176] In a possible implementation, for any immune cell receptor in the plurality of immune cell receptors, the server performs full connection on the receptor feature of the immune cell receptor through the classification module of the state prediction model to obtain a full connection matrix corresponding to the immune cell receptor. The server normalizes the full connection matrix corresponding to the immune cell receptor through the classification module of the state prediction model to output a probability set corresponding to the immune cell receptor, and the probability set includes a plurality of probabilities, each probability corresponding to a candidate state. In the case that the target object is a human and the candidate state is a disease, the probability of the plurality of candidate states is the probability that the human has the corresponding disease.

[0177] In this implementation, the server can map the receptor feature of the immune cell receptor to the probability of the plurality of candidate states through the classification module of the state prediction model, that is, the state prediction is performed at the cell receptor level, which is helpful for subsequent state prediction at the individual level.

[0178] The following will be described in combination with Figure 5 The above steps 301-305 will be described.

[0179] Referring to Figure 5For any of the plurality of immune cell receptors, the server inputs the gene information, the sequence information, and the three-dimensional structure information of the immune cell receptor into a state prediction model, the state prediction model comprising a gene encoder 501, a sequence encoder 502, and a structure encoder 503. The server encodes the gene information of the immune cell receptor through the gene encoder 501 to obtain gene features of the immune cell receptor. The server encodes the sequence information of the immune cell receptor through the sequence encoder 502 to obtain sequence features of the immune cell receptor. The server encodes the three-dimensional structure information of the immune cell receptor through the structure encoder 503 to obtain three-dimensional structure features of the immune cell receptor. The state prediction model further comprises a feature fusion module 504, and the server splices the gene features and the sequence features of the immune cell receptor through the feature fusion module 504 to obtain gene sequence fusion features h bio of the immune cell receptor. The server performs weighted fusion on the gene sequence fusion features h bio and the three-dimensional structure features h stru of the immune cell receptor based on a gated attention mechanism through the feature fusion module of the state prediction model to obtain target gene sequence fusion features h / bio and target three-dimensional structure features h / stru of the immune cell receptor. The server multiplies the target gene sequence fusion features h / bio and the target three-dimensional structure h / stru to obtain initial receptor features h fusion of the B cell receptor. The process is also referred to as a process of tensor fusion. The server performs twice full connection (FC1, FC2) on the initial receptor features h fusion through the feature fusion module 504 to obtain receptor features Representation of the B cell receptor. The state prediction model further comprises a classification module, and the server performs state prediction based on the receptor features of the immune cell receptor through the classification module of the state prediction model to determine probabilities that the immune cell receptor corresponds to a plurality of candidate states.

[0180] 306、The server determines at least one target state of the target object based on the probabilities that each immune cell receptor corresponds to a plurality of candidate states.

[0181] In a possible implementation, the server determines, based on the probability that each immune cell receptor corresponds to a plurality of candidate states, a reference candidate state corresponding to each immune cell receptor, the reference candidate state being a candidate state in the plurality of candidate states that meets a first probability condition.

[0182] The reference candidate state corresponding to each immune cell receptor is a state prediction of the cell receptor hierarchy.

[0183] In this implementation, the server can determine at least one target state of the target object according to the reference candidate state corresponding to the immune cell receptor, fully utilizes the information provided by the plurality of immune cell receptors in the immune repertoire of the target object, and the accuracy of the determined target state is high.

[0184] To make the above implementation clearer, the above implementation will be described in two parts.

[0185] In the first part, the server determines, based on the probability that each immune cell receptor corresponds to a plurality of candidate states, a reference candidate state corresponding to each immune cell receptor.

[0186] In a possible implementation, for any immune cell receptor in the plurality of immune cell receptors, the server determines, as the reference candidate state, a candidate state in the plurality of candidate states corresponding to the immune cell receptor that has a probability greater than or equal to a first probability threshold, where the first probability threshold is set by a technician according to actual conditions, and the embodiments of the present application do not limit this.

[0187] In this implementation, the server can determine the reference candidate state corresponding to each immune cell receptor through the first probability threshold, ensure that each determined reference candidate state is a candidate state with a high probability, and the accuracy of the reference candidate state is high.

[0188] In a possible implementation, for any immune cell receptor in the plurality of immune cell receptors, the server determines, as the reference candidate state, a candidate state in the plurality of candidate states corresponding to the immune cell receptor that has the highest probability.

[0189] In this implementation, the server can determine, as the reference candidate state, a candidate state in the plurality of candidate states that has the highest probability, thereby ensuring that each immune cell receptor can determine a reference candidate state, which is helpful for subsequently determining the target state of the target object.

[0190] In the second part, the server determines, based on the reference candidate state corresponding to the plurality of immune cell receptors, at least one target state of the target object.

[0191] In a possible implementation, for a first reference candidate state among the reference candidate states corresponding to the plurality of immune cell receptors, the server determines the first reference candidate state as the target state of the target object in a case where a number of times that the plurality of immune cell receptors correspond to the first reference candidate state meets a number condition.

[0192] In this implementation, the server can determine whether the first reference candidate state is the target state of the target object according to the number of occurrences of the first reference candidate state. This is a kind of "voting" mechanism, in which the plurality of immune cell receptors "vote" to determine the target state, and the accuracy of the determined target state is relatively high.

[0193] For example, for a first reference candidate state among the reference candidate states corresponding to the plurality of immune cell receptors, the server determines a number of times that the plurality of immune cell receptors correspond to the first reference candidate state. In a case where the number of times that the plurality of immune cell receptors correspond to the first reference candidate state is greater than or equal to a number threshold, the server determines the first reference candidate state as the target state of the target object, where the number threshold is set by a technician according to actual conditions, and embodiments of the present application do not limit this. In a case where the number of times that the plurality of immune cell receptors correspond to the first reference candidate state is less than the number threshold, the server does not determine the first reference candidate state as the target state of the target object. For example, the server determines whether the first reference candidate state is the target state of the target object by using the following formula (1).

[0194]

[0195] wherein g frequency (p m ) is a function of determining the target state, 0 indicates that the corresponding reference candidate state is not the target state, 1 indicates that the corresponding reference candidate state is the target state, p m is the number of occurrences of the reference candidate state numbered m, that is, the number of occurrences of the first reference candidate state, and T is the number threshold. In some embodiments, g frequency (p m ) is also referred to as a frequency-based voting mechanism.

[0196] In a possible implementation, for a second reference candidate state among the reference candidate states corresponding to the plurality of immune cell receptors, the server determines an average probability that the plurality of immune cell receptors correspond to the second reference candidate state. In a case where the average probability meets a second probability condition, the server determines the second reference candidate state as the target state of the target object.

[0197] In this implementation, the server can determine whether the second reference candidate state is the target state according to the average probability of the second reference candidate state, and the determination of the target state considers the results of multiple immune cell receptors, and the accuracy of the determination of the target state is high.

[0198] For example, for a second reference candidate state in the reference candidate states corresponding to the multiple immune cell receptors, the server determines a plurality of probabilities corresponding to the second reference candidate state. The server determines an average probability of the plurality of probabilities, which is the average probability of the multiple immune cell receptors corresponding to the second reference candidate state. In a case where the average probability is greater than or equal to a second probability threshold, the server determines the second reference candidate state as the target state of the target object, where the second probability threshold is set by a technician according to actual conditions, and embodiments of the present application do not limit this. In a case where the average probability is less than the second probability threshold, the server does not determine the second reference candidate state as the target state of the target object. For example, the server determines the average probability of the second reference candidate state by the following formula (2).

[0199]

[0200] wherein g mean (p n ) is a function of determining the target state, 0 indicates that the corresponding reference candidate state is not the target state, 1 indicates that the corresponding reference candidate state is the target state, p n is the probability of the state of the reference candidate state numbered n, that is, the second reference candidate state, p n =f(r n , θ), f(r n , θ) is a probability prediction function, r n is the reference candidate state numbered n, θ is a model parameter of a state prediction model, N is the number of the multiple immune cell receptors corresponding to the second reference candidate state, and K is the second probability threshold.

[0201] It should be noted that the above two implementations are described by taking the determination of whether the first reference candidate state and the second reference candidate state are the target state as examples, and the manner of determining whether other candidate states in the plurality of reference candidate states are the target state belongs to the same inventive concept as the above implementations, for example, the server determines at least one target state of the target object by the following formula (3).

[0202]

[0203] wherein L is the number of reference candidate states, p l is the above p m or p ng ( ) is the above g frequency ( ) or g mean ( ), is a set of at least one target state.

[0204] In a possible implementation, the server sorts the reference candidate states corresponding to the plurality of immune cell receptors in an order of probability from large to small. The server determines the first target number of reference candidate states as the target state of the target object.

[0205] The target number is set by the technical personnel according to actual conditions, for example, the first 10%, the first 25%, or the first 50% of the plurality of reference candidate states, and the present embodiment does not limit this.

[0206] In this implementation, the server can sort the reference candidate states determined based on the immune cell receptors in the order of probability, and determine the reference candidate state with a larger probability in the plurality of reference candidate states as the target state, so that the information of each immune cell receptor is fully utilized, and the accuracy of the determined target state is higher.

[0207] In a possible implementation, the server inputs the reference candidate states corresponding to the plurality of immune cell receptors and the probabilities corresponding to the reference candidate states into a state screening model. The server processes the reference candidate states corresponding to the plurality of immune cell receptors and the probabilities corresponding to the reference candidate states through the state screening model, and outputs the at least one target state.

[0208] The state screening model is trained based on a plurality of sample candidate states and probabilities corresponding to the sample candidate states, and has the ability to output a target state based on a candidate state and a probability corresponding to the candidate state.

[0209] In this implementation, the server determines the target state from the plurality of reference candidate states through the state screening model, and since the trained state screening model has a certain generalization ability, the accuracy of the determined target state is higher.

[0210] For example, the server inputs the reference candidate states corresponding to the plurality of immune cell receptors and the probabilities corresponding to the reference candidate states into a state screening model. The server convolves, fully connects, and normalizes the reference candidate states and the probabilities corresponding to the reference candidate states through the state screening model, and outputs at least one target state in the plurality of reference candidate states.

[0211] For example, the server determines a state determination matrix composed of the reference candidate states corresponding to the plurality of immune cell receptors and the probabilities corresponding to the reference candidate states. The server inputs the state determination matrix into the state screening model, and performs convolution, full connection and normalization on the state determination matrix through the state screening model to output a probability set corresponding to the state screening matrix, the probability set including a plurality of probabilities, each probability corresponding to a candidate state. The server determines the reference candidate state corresponding to the probability greater than or equal to the third probability threshold in the probability set as the target state.

[0212] The following will be described in combination with Figure 6 The above steps 301-306 are described.

[0213] Referring to Figure 6 The server determines a plurality of immune cell receptors 602 in the immune repertoire of the target object 601. The server obtains the gene information 6031, the sequence information 6032 and the three-dimensional structure information 6033 of the plurality of immune cell receptors. The server inputs the gene information 6031, the sequence information 6032 and the three-dimensional structure information 6033 of the plurality of immune cell receptors into the state prediction model 604, and outputs the probability 605 of each immune cell receptor corresponding to a plurality of candidate states through the state prediction model 604. The server determines at least one target state 606 of the target object based on the probability 605 of each immune cell receptor corresponding to a plurality of candidate states.

[0214] In a possible implementation, the state prediction method provided by the embodiments of the present application can be provided as a cloud service, that is, the user transmits relevant data to the cloud server through the terminal, the cloud server executes the state prediction method provided by the embodiments of the present application, and returns at least one target state to the terminal, and the terminal displays the at least one target state to the user.

[0215] For example, referring to Figure 7The terminal displays a state prediction interface 700, which is used to acquire gene information, sequence information and three-dimensional structure characteristics of a plurality of immune cell receptors in an immune repertoire of a target object. The state prediction interface 700 includes a file uploading control 701, by which a file carrying the above information can be uploaded. The state prediction interface 700 further includes a state selection area 702, which is used to select a state that is intended to be predicted. In the case where the state is a disease, the state selection area 702 is also a disease selection area. The state prediction interface 700 further includes a submission control 703. In response to a click operation on the submission control 703, the terminal uploads the gene information, sequence information and three-dimensional structure characteristics of the plurality of immune cell receptors to a cloud server, and the cloud server performs calculation based on the gene information, sequence information and three-dimensional structure characteristics of the plurality of immune cell receptors to obtain whether the target object is in the state selected in the state selection area 702. For example, referring to Figure 8 The terminal displays a state display interface 800, in which whether the target object is in the state selected in the state selection area 702 and a probability of being in the state are displayed.

[0216] All the optional technical solutions described above can be combined to form optional embodiments of the present application, which will not be described again.

[0217] Figure 9 The prediction results of the technical solutions provided in the embodiments of the present application (prediction of whether a patient has inflammatory bowel disease (IBD) based on a BCR immune repertoire) are shown. Figure 9 A shows the readiness of single receptor level prediction for each patient. Figure 9 B shows the probability distribution of receptor level prediction for each patient. Figure 9 C and D show the results of using top 10%, top 25%, average value based and frequency of receptors with a prediction probability greater than 0.5 to perform repertoire level characterization for each patient after fusion. Figure 10 The results of the technical solutions provided in the embodiments of the present application in predicting whether a patient has inflammatory bowel disease (IBD) based on a TCR immune repertoire are shown. Figure 10 A shows the readiness of single receptor level prediction for each patient. Figure 10 B shows the probability distribution of receptor level prediction for each patient. Figure 10 C and D show the results of using top 10%, top 25%, average value based and frequency of receptors with a prediction probability greater than 0.5 to perform repertoire level characterization for each patient after fusion. Figure 9 and Figure 10It can be seen that the technical solution provided in the embodiments of the present application can realize correct prediction at the repertoire level. The repertoire level refers to the individual level, that is, the level of predicting the state of the target object. The receptor level is the level predicted based on the immune cell receptor.

[0218] Through the technical solution provided in the embodiments of the present application, the state prediction model extracts features from the gene information and sequence of the immune cell receptor to obtain the gene features and sequence features of the immune cell receptor. In the process of obtaining the receptor features of the immune cell receptor, the gene features, sequence features and three-dimensional structure features are fused. The introduction of the three-dimensional structure features enriches the content of the receptor features and improves the expression ability of the receptor features, so that the accuracy of the state prediction of the target object based on the receptor features is high.

[0219] The technical solution provided in the embodiments of the present application can perform related prediction at the individual level based on the entire immune repertoire information and give the association of each BCR / TCR and disease. Because the immune repertoire contains the immune response of the individual to various diseases and cancers, the technical solution provided in the embodiments of the present application helps to explore whether the state of multiple diseases can be inferred at the same time through one immune repertoire detection. In addition, the technical solution provided in the embodiments of the present application can indicate the TCR and BCR related to the predicted state of the individual. Therefore, it can screen and provide effective TCR / BCR for biological drug research and development and clinical application. The technical solution provided in the embodiments of the present application can utilize massive biological data to optimize drug development, improve the efficiency of new drug research and development, and provide a basis for precision medicine, immunotherapy and vaccine development.

[0220] In order to more clearly illustrate the state prediction method provided in the embodiments of the present application, the training method of the state prediction model provided in the embodiments of the present application is described below, referring to Figure 11 Taking a server as an execution subject, the method includes the following steps.

[0221] 1101. The server inputs the gene information, sequence information and three-dimensional structure features of a plurality of immune cell receptors in the immune repertoire of the sample object into the state prediction model.

[0222] Step 1101 and step 302 described above belong to the same inventive concept, and the implementation process is described above with reference to the related description of step 302, which will not be described here.

[0223] 1102. The server extracts features from the gene information and sequence information of the plurality of immune cell receptors through the state prediction model to obtain the gene features and sequence features of each immune cell receptor.

[0224] Step 1102 belongs to the same inventive concept as step 303 described above, and the implementation process is described above in relation to step 303, which will not be repeated here.

[0225] 1103. The server fuses the gene features, sequence features, and three-dimensional structure features of each immune cell receptor by using the state prediction model to obtain receptor features of each immune cell receptor.

[0226] Step 1103 belongs to the same inventive concept as step 304 described above, and the implementation process is described above in relation to step 304, which will not be repeated here.

[0227] 1104. The server performs state prediction based on the receptor features of each immune cell receptor by using the state prediction model, and outputs at least one sample state of the sample object.

[0228] Step 1104 belongs to the same inventive concept as steps 305 and 306 described above, and the implementation process is described above in relation to steps 305 and 306, which will not be repeated here.

[0229] 1105. The server trains the state prediction model based on the difference information between the at least one sample state of the sample object and the labeled state.

[0230] In one possible implementation, the server constructs a cross-entropy loss function based on the difference information between the sample state of the sample object and the labeled state. The server uses the gradient descent method to train the state prediction model using the cross-entropy loss function, that is, to adjust the model parameters of the state prediction model.

[0231] It should be noted that steps 1101-1105 above are described by taking one round of training of the state prediction model by the server as an example. The process of multiple rounds of training of the state prediction model belongs to the same inventive concept as steps 1101-1105 described above, which will not be repeated here.

[0232] Figure 12 is a structural schematic diagram of a state prediction device provided by an embodiment of the present application, referring to Figure 12 The device comprises an input unit 1201, a feature extraction unit 1202, a feature fusion unit 1203, and a state prediction unit 1204.

[0233] The input unit 1201 is configured to input gene information, sequence information, and three-dimensional structure features of a plurality of immune cell receptors in an immune repertoire of a target object into a state prediction model.

[0234] The feature extraction unit 1202 is configured to perform feature extraction on the gene information and the sequence information of the plurality of immune cell receptors by using the state prediction model, to obtain gene features and sequence features of each of the immune cell receptors.

[0235] The feature fusion unit 1203 is configured to fuse the gene features, the sequence features, and the three-dimensional structure features of each of the immune cell receptors by using the state prediction model, to obtain receptor features of each of the immune cell receptors.

[0236] The state prediction unit 1204 is configured to perform state prediction based on the receptor features of each of the immune cell receptors by using the state prediction model, and output at least one target state of the target object.

[0237] In a possible implementation, the feature extraction unit 1202 is configured to, for any one of the plurality of immune cell receptors, encode VDJ information of the immune cell receptor by using a gene encoder of the state prediction model, to obtain gene features of the immune cell receptor, where V represents a variable region, D represents a hypervariable region, and J represents a junction region; and encode an amino acid sequence of the immune cell receptor by using a sequence encoder of the state prediction model, to obtain sequence features of the immune cell receptor.

[0238] In a possible implementation, the feature extraction unit 1202 is configured to perform any one of the following:

[0239] In a case where the immune cell receptor is a B cell receptor, the VJ information of a light chain and the VDJ information of a heavy chain of the immune cell receptor are encoded to obtain the gene features of the immune cell receptor.

[0240] In a case where the immune cell receptor is a T cell receptor, the VJ information of an alpha chain and the VDJ information of a beta chain of the immune cell receptor are encoded to obtain the gene features of the immune cell receptor.

[0241] In a possible implementation, the feature extraction unit 1202 is configured to perform full connection on the VJ information of the light chain and the VDJ information of the heavy chain of the immune cell receptor, to obtain the gene features of the immune cell receptor, where the gene features of the immune cell receptor include light chain gene features of the immune cell receptor and heavy chain gene features of the immune cell receptor. The VJ information of the alpha chain and the VDJ information of the beta chain of the immune cell receptor are encoded to obtain the gene features of the immune cell receptor, including performing full connection on the VJ information of the alpha chain and the VDJ information of the beta chain of the immune cell receptor, to obtain the gene features of the immune cell receptor, where the gene features of the immune cell receptor include alpha chain gene features of the immune cell receptor and beta chain gene features of the immune cell receptor.

[0242] In a possible implementation, the feature extraction unit 1202 is configured to perform any one of the following:

[0243] In the case where the immune cell receptor is a B cell receptor, the sequence encoder of the state prediction model encodes the amino acid sequence of the light chain and the amino acid sequence of the heavy chain of the immune cell receptor based on an attention mechanism to obtain sequence features of the immune cell receptor, the sequence features of the immune cell receptor including light chain sequence features and heavy chain sequence features of the immune cell receptor.

[0244] In the case where the immune cell receptor is a T cell receptor, the sequence encoder of the state prediction model encodes the amino acid sequence of the alpha chain and the amino acid sequence of the beta chain of the immune cell receptor based on an attention mechanism to obtain sequence features of the immune cell receptor, the sequence features of the immune cell receptor including alpha chain sequence features and beta chain sequence features of the immune cell receptor.

[0245] In a possible implementation, the feature fusion unit 1203 is configured to, for any immune cell receptor in the plurality of immune cell receptors, splice the gene features and the sequence features of the immune cell receptor by the feature fusion of the state prediction model to obtain gene sequence fusion features of the immune cell receptor.

[0246] The gene sequence fusion features and the three-dimensional structure features of the immune cell receptor are weighted and fused based on a gated attention mechanism to obtain receptor features of the immune cell receptor.

[0247] In a possible implementation, the state prediction unit 1204 is configured to perform full connection and normalization on the receptor features of each immune cell receptor by the classification module of the state prediction model to output probabilities of each immune cell receptor corresponding to a plurality of candidate states. At least one target state of the target object is determined based on the probabilities of each immune cell receptor corresponding to the plurality of candidate states.

[0248] In a possible implementation, the state prediction unit 1204 is configured to determine a reference candidate state corresponding to each immune cell receptor based on the probabilities of each immune cell receptor corresponding to a plurality of candidate states, the reference candidate state being a candidate state in the plurality of candidate states that meets a first probability condition. At least one target state of the target object is determined based on the reference candidate states corresponding to the plurality of immune cell receptors.

[0249] In a possible implementation, the state prediction unit 1204 is configured to perform any one of the following:

[0250] For a first reference candidate state of the reference candidate states corresponding to the plurality of immune cell receptors, the first reference candidate state is determined as the target state of the target object in a case where a number of times that the plurality of immune cell receptors correspond to the first reference candidate state meets a number condition.

[0251] For a second reference candidate state of the reference candidate states corresponding to the plurality of immune cell receptors, an average probability that the plurality of immune cell receptors correspond to the second reference candidate state is determined. The second reference candidate state is determined as the target state of the target object in a case where the average probability meets a second probability condition.

[0252] In a possible implementation, the state prediction unit 1204 is configured to perform any one of the following:

[0253] The reference candidate states corresponding to the plurality of immune cell receptors are sorted in an order from large to small according to the probabilities. The first target number of reference candidate states are determined as the target states of the target object.

[0254] The reference candidate states corresponding to the plurality of immune cell receptors and the probabilities corresponding to the reference candidate states are input into a state screening model. The reference candidate states corresponding to the plurality of immune cell receptors and the probabilities corresponding to the reference candidate states are processed by the state screening model, and the at least one target state is output.

[0255] In a possible implementation, the apparatus further includes a three-dimensional structure feature determination unit configured to, for any immune cell receptor of the plurality of immune cell receptors, obtain an amino acid sequence of a CDR3 region of the immune cell receptor. The amino acid sequence of the CDR3 region of the immune cell receptor is subjected to multiple sequence alignment to obtain at least one reference amino acid sequence, similarity between the reference amino acid sequence and the amino acid sequence of the CDR3 region of the immune cell receptor meeting a similarity condition. A homology template corresponding to the amino acid sequence of the CDR3 region of the immune cell receptor is obtained, the homology template including structure information of a homologous sequence of the amino acid sequence of the CDR3 region of the immune cell receptor. The amino acid sequence of the CDR3 region of the immune cell receptor, the at least one reference amino acid sequence, and the homology template are subjected to multiple rounds of iteration to obtain a three-dimensional structure feature of the immune cell receptor.

[0256] In a possible implementation, the apparatus further includes a three-dimensional structure feature determination unit configured to, for any immune cell receptor of the plurality of immune cell receptors, obtain three-dimensional structure information of the immune cell receptor, the three-dimensional structure information including three-dimensional coordinates of a plurality of amino acids in the immune cell receptor.

[0257] The three-dimensional structure feature determination unit is further configured to perform any one of the following:

[0258] The three-dimensional structure information of the immune cell receptor is graph-convolved to obtain a three-dimensional structure feature of the immune cell receptor.

[0259] The three-dimensional structure information of the immune cell receptor is encoded based on an attention mechanism to obtain a three-dimensional structure feature of the immune cell receptor.

[0260] In a possible implementation, the feature fusion unit 1203 is further configured to fuse, by the state prediction model, the gene feature, the sequence feature, the three-dimensional structure feature of each immune cell receptor, and the physicochemical information of the amino acid in each immune cell receptor to obtain a receptor feature of each immune cell receptor.

[0261] It should be noted that the state prediction apparatus provided in the above embodiments is only used as an example to illustrate the division of the above functional modules in state prediction. In actual applications, the above functions can be completed by different functional modules according to needs, that is, the internal structure of the computer device is divided into different functional modules to complete all or part of the functions described above. In addition, the state prediction apparatus and the state prediction method embodiments provided in the above embodiments belong to the same concept, and the specific implementation process is described in detail in the method embodiments, which will not be described here.

[0262] Through the technical solutions provided in the embodiments of the present application, the state prediction model extracts features from the gene information and the sequence of the immune cell receptor to obtain the gene feature and the sequence feature of the immune cell receptor. In the process of obtaining the receptor feature of the immune cell receptor, the gene feature, the sequence feature, and the three-dimensional structure feature are fused. The introduction of the three-dimensional structure feature enriches the content of the receptor feature and improves the expression ability of the receptor feature, so that the accuracy of the state prediction of the target object based on the receptor feature is higher.

[0263] Figure 13 is a structural schematic diagram of a state prediction model training apparatus provided in an embodiment of the present application, referring to Figure 13 The apparatus comprises a training information input unit 1301, a training feature extraction unit 1302, a training feature fusion unit 1303, a sample state output unit 1304, and a training unit 1305.

[0264] The training information input unit 1301 is configured to input the gene information, the sequence information, and the three-dimensional structure feature of a plurality of immune cell receptors in the immune repertoire of a sample object into the state prediction model.

[0265] The training feature extraction unit 1302 is configured to extract features from the gene information and the sequence information of the plurality of immune cell receptors by the state prediction model to obtain the gene feature and the sequence feature of each immune cell receptor.

[0266] The training feature fusion unit 1303 is configured to fuse the gene features, the sequence features and the three-dimensional structure features of each immune cell receptor to obtain receptor features of each immune cell receptor by using the state prediction model.

[0267] The sample state output unit 1304 is configured to perform state prediction based on the receptor features of each immune cell receptor by using the state prediction model, and output at least one sample state of the sample object.

[0268] The training unit 1305 is configured to train the state prediction model based on difference information between the at least one sample state of the sample object and the labeled state.

[0269] It should be noted that the training device of the state prediction model provided in the above embodiments is only used as an example to illustrate the division of the above functional modules in training the state prediction model. In actual applications, the above functions can be completed by different functional modules according to needs, that is, the internal structure of the computer device is divided into different functional modules to complete all or part of the functions described above. In addition, the state prediction device and the state prediction method provided in the above embodiments belong to the same concept, and the specific implementation process is described in the method embodiments, which will not be repeated here.

[0270] The embodiments of the present application provide a computer device for executing the above method. The computer device can be implemented as a terminal or a server. First, the structure of the terminal will be introduced as follows:

[0271] Figure 14 FIG. 1 is a structural schematic diagram of a terminal provided in an embodiment of the present application. The terminal 1400 can be a smart phone, a tablet computer, a notebook computer or a desktop computer. The terminal 1400 can also be referred to as a user equipment, a portable terminal, a laptop terminal, a desktop terminal or other names.

[0272] Generally, the terminal 1400 includes one or more processors 1401 and one or more memories 1402.

[0273] The processor 1401 can include one or more processing cores, such as a 4-core processor, an 8-core processor, and the like. The processor 1401 can be implemented in at least one of a hardware form of a DSP (Digital Signal Processing), an FPGA (Field-Programmable Gate Array), a PLA (Programmable Logic Array). The processor 1401 can also include a main processor and a coprocessor, the main processor being a processor for processing data in an awake state, also known as a CPU (Central Processing Unit), and the coprocessor being a low-power processor for processing data in a standby state. In some embodiments, the processor 1401 can be integrated with a GPU (Graphics Processing Unit) that is responsible for rendering and drawing the content required to be displayed by the display screen. In some embodiments, the processor 1401 can also include an AI (Artificial Intelligence) processor for processing computing operations related to machine learning.

[0274] The memory 1402 can include one or more computer-readable storage media that can be non-transitory. The memory 1402 can also include a high-speed random access memory, and a nonvolatile memory such as one or more disk storage devices, flash storage devices. In some embodiments, the non-transitory computer-readable storage medium in the memory 1402 is used to store at least one computer program for being executed by the processor 1401 to implement the state prediction method or the training method of the state prediction model provided by the method embodiments in the present application.

[0275] In some embodiments, the terminal 1400 can also optionally include a peripheral device interface 1403 and at least one peripheral device. The processor 1401, the memory 1402, and the peripheral device interface 1403 can be connected through a bus or a signal line. Each peripheral device can be connected to the peripheral device interface 1403 through a bus, a signal line, or a circuit board. Specifically, the peripheral device includes at least one of a radio frequency circuit 1404, a display screen 1405, a camera assembly 1406, an audio circuit 1407, and a power supply 1408.

[0276] The peripheral interface 1403 can be used to connect at least one I / O (Input / Output) related peripheral device to the processor 1401 and the memory 1402. In some embodiments, the processor 1401, the memory 1402 and the peripheral interface 1403 are integrated on the same chip or circuit board; in some other embodiments, any one or two of the processor 1401, the memory 1402 and the peripheral interface 1403 can be implemented on a separate chip or circuit board, and the present embodiments are not limited to this.

[0277] The radio frequency circuit 1404 is used to receive and send RF (Radio Frequency) signals, also known as electromagnetic signals. The radio frequency circuit 1404 communicates with communication networks and other communication devices through electromagnetic signals. The radio frequency circuit 1404 converts electrical signals into electromagnetic signals for transmission, or converts received electromagnetic signals into electrical signals. Optionally, the radio frequency circuit 1404 includes an antenna system, an RF transceiver, one or more amplifiers, a tuner, an oscillator, a digital signal processor, a codec chipset, a subscriber identity module card, and the like.

[0278] The display screen 1405 is used to display a UI (User Interface). The UI can include graphics, text, icons, video, and any combination thereof. When the display screen 1405 is a touch display screen, the display screen 1405 also has the ability to collect touch signals on or above the surface of the display screen 1405. The touch signals can be input as control signals to the processor 1401 for processing. At this time, the display screen 1405 can also be used to provide virtual buttons and / or virtual keyboards, also known as soft buttons and / or soft keyboards.

[0279] The camera assembly 1406 is used to capture images or videos. Optionally, the camera assembly 1406 includes a front-facing camera and a rear-facing camera. Typically, the front-facing camera is disposed on the front panel of the terminal, and the rear-facing camera is disposed on the back of the terminal.

[0280] The audio circuit 1407 can include a microphone and a speaker. The microphone is used to collect sound waves of the user and the environment, and convert the sound waves into electrical signals input to the processor 1401 for processing, or input to the radio frequency circuit 1404 to realize voice communication.

[0281] The power supply 1408 is used to supply power to various components in the terminal 1400. The power supply 1408 can be alternating current, direct current, disposable batteries or rechargeable batteries.

[0282] In some embodiments, the terminal 1400 further comprises one or more sensors 1409. The one or more sensors 1409 include, but are not limited to, an acceleration sensor 1410, a gyroscope sensor 1411, a pressure sensor 1412, an optical sensor 1413, and a proximity sensor 1414.

[0283] The acceleration sensor 1410 can detect the acceleration magnitude in three coordinate axes of a coordinate system established by the terminal 1400.

[0284] The gyroscope sensor 1411 can detect the body direction and rotation angle of the terminal 1400. The gyroscope sensor 1411 can cooperate with the acceleration sensor 1410 to collect the 3D action of the user on the terminal 1400.

[0285] The pressure sensor 1412 can be arranged on the side frame of the terminal 1400 and / or the lower layer of the display screen 1405. When the pressure sensor 1412 is arranged on the side frame of the terminal 1400, the holding signal of the user on the terminal 1400 can be detected, and the left-hand or right-hand recognition or shortcut operation can be performed by the processor 1401 according to the holding signal collected by the pressure sensor 1412. When the pressure sensor 1412 is arranged on the lower layer of the display screen 1405, the controllable control on the UI interface can be controlled by the processor 1401 according to the pressure operation of the user on the display screen 1405.

[0286] The optical sensor 1413 is used to collect the ambient light intensity. In one embodiment, the processor 1401 can control the display brightness of the display screen 1405 according to the ambient light intensity collected by the optical sensor 1413.

[0287] The proximity sensor 1414 is used to collect the distance between the user and the front of the terminal 1400.

[0288] Those skilled in the art can understand that, Figure 14 The structure shown in the above description does not constitute a limitation on the terminal 1400, and can include more or less components than the illustrated components, or combine certain components, or adopt different component arrangements.

[0289] The above computer device can also be implemented as a server, and the structure of the server will be introduced as follows:

[0290] Figure 15is a structural schematic diagram of a server provided by an embodiment of the present application. The server 1500 can have great differences due to different configurations or performances, and can include one or more processors (Central Processing Units, CPUs) 1501 and one or more memories 1502. The one or more memories 1502 store at least one computer program, which is loaded and executed by the one or more processors 1501 to implement the method provided by each of the above methods. Of course, the server 1500 can also have a wired or wireless network interface, a keyboard, an input and output interface, and other components for realizing the functions of the device, and the like, so as to perform input and output. The server 1500 can also include other components for realizing the functions of the device, which are not described herein.

[0291] In an example embodiment, a computer-readable storage medium, such as a memory including a computer program, is also provided. The computer program can be executed by a processor to complete the state prediction method or the training method of the state prediction model in the above embodiments. For example, the computer-readable storage medium can be a Read-Only Memory (ROM), a Random Access Memory (RAM), a Compact Disc Read-Only Memory (CD-ROM), a magnetic tape, a floppy disk, and an optical data storage device, etc.

[0292] In an example embodiment, a computer program product or computer program is also provided. The computer program product or computer program includes program code stored in a computer-readable storage medium. The processor of a computer device reads the program code from the computer-readable storage medium. The processor executes the program code to make the computer device execute the state prediction method or the training method of the state prediction model.

[0293] In some embodiments, the computer program related to the embodiments of the present application can be deployed on a computer device for execution, or on multiple computer devices located in one place, or on multiple computer devices distributed in multiple places and interconnected through a communication network. The multiple computer devices distributed in multiple places and interconnected through a communication network can constitute a blockchain system.

[0294] Those of ordinary skill in the art can understand that all or part of the steps of the above embodiments can be completed by hardware, or by a program instructing relevant hardware, which can be stored in a computer-readable storage medium. The storage medium mentioned above can be a Read-Only Memory, a magnetic disk or an optical disk, etc.

[0295] The above merely is the optional embodiment of the present application, and does not use to limit the present application, any modification, equivalent replacement, improvement and so on, which are made in the spirit and principle of the present application, should be included in the protection scope of the present application.

Claims

1. A state prediction method characterized by, The method comprises: For any one of a plurality of immune cell receptors in an immune repertoire of a target object, obtain three-dimensional structure information of the immune cell receptor, the three-dimensional structure information comprising three-dimensional coordinates of a plurality of amino acids in the immune cell receptor; Perform graph convolution on the three-dimensional structure information of the immune cell receptor to obtain a three-dimensional structure feature of the immune cell receptor; or, encode the three-dimensional structure information of the immune cell receptor based on an attention mechanism to obtain the three-dimensional structure feature of the immune cell receptor; Input the gene information, sequence information and three-dimensional structure feature of the plurality of immune cell receptors into a state prediction model; Extract features of the gene information and sequence information of the plurality of immune cell receptors by the state prediction model to obtain a gene feature and a sequence feature of each of the immune cell receptors; Fuse the gene feature, sequence feature and three-dimensional structure feature of each of the immune cell receptors by the state prediction model to obtain a receptor feature of each of the immune cell receptors; Perform state prediction based on the receptor feature of each of the immune cell receptors by the state prediction model to output at least one target state of the target object.

2. The method of claim 1, wherein, The feature extraction of the gene information and sequence information of the plurality of immune cell receptors by the state prediction model to obtain the gene feature and sequence feature of each of the immune cell receptors comprises: For any one of the plurality of immune cell receptors, encode VDJ information of the immune cell receptor by a gene encoder of the state prediction model to obtain a gene feature of the immune cell receptor, wherein V is a variable region, D is a hypervariable region, and J is a cross-linking region; Encode an amino acid sequence of the immune cell receptor by a sequence encoder of the state prediction model to obtain a sequence feature of the immune cell receptor.

3. The method of claim 2, wherein, The encoding of the VDJ information of the immune cell receptor to obtain the gene feature of the immune cell receptor comprises any one of the following: In the case where the immune cell receptor is a B cell receptor, encode VJ information of a light chain and VDJ information of a heavy chain of the immune cell receptor to obtain the gene feature of the immune cell receptor; In the case where the immune cell receptor is a T cell receptor, encode VJ information of an alpha chain and VDJ information of a beta chain of the immune cell receptor to obtain the gene feature of the immune cell receptor.

4. The method of claim 3, wherein, The encoding of the VJ information of the light chain and the VDJ information of the heavy chain of the immune cell receptor to obtain the gene feature of the immune cell receptor comprises: Fully connect the VJ information of the light chain and the VDJ information of the heavy chain of the immune cell receptor to obtain the gene feature of the immune cell receptor, the gene feature of the immune cell receptor comprising a light chain gene feature of the immune cell receptor and a heavy chain gene feature of the immune cell receptor; The encoding of the VJ information of the alpha chain and the VDJ information of the beta chain of the immune cell receptor to obtain the gene feature of the immune cell receptor comprises: The VJ information of the alpha chain and the VDJ information of the beta chain of the immune cell receptor are fully connected to obtain a gene feature of the immune cell receptor, and the gene feature of the immune cell receptor includes an alpha chain gene feature of the immune cell receptor and a beta chain gene feature of the immune cell receptor.

5. The method of claim 2, wherein, The sequence encoder of the state prediction model encodes the amino acid sequence of the immune cell receptor to obtain a sequence feature of the immune cell receptor, and the sequence feature of the immune cell receptor includes any one of the following: In a case where the immune cell receptor is a B cell receptor, the sequence encoder of the state prediction model encodes the amino acid sequence of the light chain and the amino acid sequence of the heavy chain of the immune cell receptor based on an attention mechanism to obtain a sequence feature of the immune cell receptor, and the sequence feature of the immune cell receptor includes a light chain sequence feature and a heavy chain sequence feature of the immune cell receptor. In a case where the immune cell receptor is a T cell receptor, the sequence encoder of the state prediction model encodes the amino acid sequence of the alpha chain and the amino acid sequence of the beta chain of the immune cell receptor based on an attention mechanism to obtain a sequence feature of the immune cell receptor, and the sequence feature of the immune cell receptor includes an alpha chain sequence feature and a beta chain sequence feature of the immune cell receptor.

6. The method of claim 1, wherein, The state prediction model fuses the gene feature, the sequence feature, and the three-dimensional structure feature of each immune cell receptor to obtain a receptor feature of each immune cell receptor, and the receptor feature of each immune cell receptor includes any one of the following: For any immune cell receptor in the plurality of immune cell receptors, the gene feature and the sequence feature of the immune cell receptor are spliced by the feature fusioner of the state prediction model to obtain a gene sequence fusion feature of the immune cell receptor. The gene sequence fusion feature and the three-dimensional structure feature of the immune cell receptor are weighted and fused based on a gated attention mechanism to obtain a receptor feature of the immune cell receptor.

7. The method of claim 1, wherein, The state prediction model predicts the state based on the receptor feature of each immune cell receptor to output at least one target state of the target object, and the output of at least one target state of the target object includes the following: The classification module of the state prediction model fully connects and normalizes the receptor feature of each immune cell receptor to output a probability of each immune cell receptor corresponding to a plurality of candidate states. Based on the probability of each immune cell receptor corresponding to a plurality of candidate states, at least one target state of the target object is determined.

8. The method of claim 7, wherein, Based on the probability of each immune cell receptor corresponding to a plurality of candidate states, a reference candidate state corresponding to each immune cell receptor is determined, and the reference candidate state is a candidate state in the plurality of candidate states that meets a first probability condition; Based on the reference candidate state corresponding to the plurality of immune cell receptors, at least one target state of the target object is determined. ​ 9. The method of claim 8, wherein, The determining the at least one target state of the target object based on the reference candidate states corresponding to the plurality of immune cell receptors comprises any of the following: For a first reference candidate state of the reference candidate states corresponding to the plurality of immune cell receptors, in a case where a number of times that the plurality of immune cell receptors correspond to the first reference candidate state meets a number condition, the first reference candidate state is determined as a target state of the target object; For a second reference candidate state of the reference candidate states corresponding to the plurality of immune cell receptors, an average probability that the plurality of immune cell receptors correspond to the second reference candidate state is determined; In a case where the average probability meets a second probability condition, the second reference candidate state is determined as the target state of the target object.

10. The method of claim 8, wherein, The determining the at least one target state of the target object based on the reference candidate states corresponding to the plurality of immune cell receptors comprises any of the following: The reference candidate states corresponding to the plurality of immune cell receptors are sorted in an order from large to small according to the probabilities, and a front target number of reference candidate states are determined as the target states of the target object; The reference candidate states corresponding to the plurality of immune cell receptors and the probabilities corresponding to the reference candidate states are input into a state screening model, and the reference candidate states corresponding to the plurality of immune cell receptors and the probabilities corresponding to the reference candidate states are processed by the state screening model to output the at least one target state.

11. The method of claim 1, wherein, Before the gene information, the sequence information, and the three-dimensional structure feature of the plurality of immune cell receptors in the immune repertoire of the target object are input into the state prediction model, the method comprises: For any immune cell receptor of the plurality of immune cell receptors, an amino acid sequence of a CDR3 region of the immune cell receptor is obtained; The amino acid sequence of the CDR3 region of the immune cell receptor is subjected to multiple sequence alignment to obtain at least one reference amino acid sequence, and a similarity between the reference amino acid sequence and the amino acid sequence of the CDR3 region of the immune cell receptor meets a similarity condition; A homologous template corresponding to the amino acid sequence of the CDR3 region of the immune cell receptor is obtained, and the homologous template comprises structure information of a homologous sequence of the amino acid sequence of the CDR3 region of the immune cell receptor; The gene information, the sequence information, and the three-dimensional structure feature of the plurality of immune cell receptors are input into the state prediction model, and the gene information and the sequence information of the plurality of immune cell receptors are subjected to feature extraction by the state prediction model to obtain gene features and sequence features of each of the plurality of immune cell receptors.

12. The method of claim 1, wherein, After the gene information and the sequence information of the plurality of immune cell receptors are subjected to feature extraction by the state prediction model, the method further comprises: The gene features, the sequence features, the three-dimensional structure features, and physicochemical information of amino acids in each of the plurality of immune cell receptors are fused by the state prediction model to obtain receptor features of each of the plurality of immune cell receptors.

13. A training method for a state prediction model, characterized in that, The method comprises: For any one of a plurality of immune cell receptors in an immune repertoire of a sample object, three-dimensional structure information of the immune cell receptor is acquired, the three-dimensional structure information comprising three-dimensional coordinates of a plurality of amino acids in the immune cell receptor; graph convolution is performed on the three-dimensional structure information of the immune cell receptor to obtain three-dimensional structure features of the immune cell receptor, or the three-dimensional structure information of the immune cell receptor is encoded based on an attention mechanism to obtain the three-dimensional structure features of the immune cell receptor; gene information, sequence information, and three-dimensional structure features of the plurality of immune cell receptors are input into a state prediction model; gene features and sequence features of each of the immune cell receptors are obtained by performing feature extraction on the gene information and the sequence information of the plurality of immune cell receptors through the state prediction model; receptor features of each of the immune cell receptors are obtained by fusing the gene features, the sequence features, and the three-dimensional structure features of each of the immune cell receptors through the state prediction model; at least one sample state of the sample object is output by performing state prediction based on the receptor features of each of the immune cell receptors through the state prediction model; the state prediction model is trained based on difference information between the at least one sample state of the sample object and a labeled state.

14. A state prediction device characterized by comprising: The device comprises: a three-dimensional structure feature determination unit configured to, for any one of a plurality of immune cell receptors in an immune repertoire of a target object, acquire three-dimensional structure information of the immune cell receptor, the three-dimensional structure information comprising three-dimensional coordinates of a plurality of amino acids in the immune cell receptor; the three-dimensional structure feature determination unit is further configured to perform graph convolution on the three-dimensional structure information of the immune cell receptor to obtain three-dimensional structure features of the immune cell receptor, or encode the three-dimensional structure information of the immune cell receptor based on an attention mechanism to obtain the three-dimensional structure features of the immune cell receptor; an input unit configured to input gene information, sequence information, and three-dimensional structure features of the plurality of immune cell receptors into a state prediction model; a feature extraction unit configured to obtain gene features and sequence features of each of the immune cell receptors by performing feature extraction on the gene information and the sequence information of the plurality of immune cell receptors through the state prediction model; a feature fusion unit configured to obtain receptor features of each of the immune cell receptors by fusing the gene features, the sequence features, and the three-dimensional structure features of each of the immune cell receptors through the state prediction model; a state prediction unit configured to output at least one target state of the target object by performing state prediction based on the receptor features of each of the immune cell receptors through the state prediction model.

15. The apparatus of claim 14, wherein, the feature extraction comprises: The VDJ information of the immune cell receptor is encoded by a gene encoder of the state prediction model to obtain a gene feature of the immune cell receptor, where V represents a variable region, D represents a hypervariable region, and J represents a junction region. The amino acid sequence of the immune cell receptor is encoded by a sequence encoder of the state prediction model to obtain a sequence feature of the immune cell receptor.

16. The apparatus of claim 15, wherein, The encoding of the VDJ information of the immune cell receptor to obtain the gene feature of the immune cell receptor includes any one of the following: In a case where the immune cell receptor is a B cell receptor, the VJ information of a light chain and the VDJ information of a heavy chain of the immune cell receptor are encoded to obtain the gene feature of the immune cell receptor. In a case where the immune cell receptor is a T cell receptor, the VJ information of an alpha chain and the VDJ information of a beta chain of the immune cell receptor are encoded to obtain the gene feature of the immune cell receptor.

17. The apparatus of claim 16, wherein, The encoding of the VJ information of the light chain and the VDJ information of the heavy chain of the immune cell receptor to obtain the gene feature of the immune cell receptor includes: The VJ information of the light chain and the VDJ information of the heavy chain of the immune cell receptor are fully connected to obtain the gene feature of the immune cell receptor, and the gene feature of the immune cell receptor includes a light chain gene feature of the immune cell receptor and a heavy chain gene feature of the immune cell receptor. The encoding of the VJ information of the alpha chain and the VDJ information of the beta chain of the immune cell receptor to obtain the gene feature of the immune cell receptor includes: The VJ information of the alpha chain and the VDJ information of the beta chain of the immune cell receptor are fully connected to obtain the gene feature of the immune cell receptor, and the gene feature of the immune cell receptor includes an alpha chain gene feature of the immune cell receptor and a beta chain gene feature of the immune cell receptor.

18. The apparatus of claim 15, wherein, The encoding of the amino acid sequence of the immune cell receptor by the sequence encoder of the state prediction model to obtain the sequence feature of the immune cell receptor includes any one of the following: In a case where the immune cell receptor is a B cell receptor, the amino acid sequence of the light chain and the amino acid sequence of the heavy chain of the immune cell receptor are encoded by the sequence encoder of the state prediction model based on an attention mechanism to obtain the sequence feature of the immune cell receptor, and the sequence feature of the immune cell receptor includes a light chain sequence feature and a heavy chain sequence feature of the immune cell receptor. In a case where the immune cell receptor is a T cell receptor, the amino acid sequence of the alpha chain and the amino acid sequence of the beta chain of the immune cell receptor are encoded by the sequence encoder of the state prediction model based on an attention mechanism to obtain the sequence feature of the immune cell receptor, and the sequence feature of the immune cell receptor includes an alpha chain sequence feature and a beta chain sequence feature of the immune cell receptor.

19. The apparatus of claim 14, wherein, The gene feature, the sequence feature, and the three-dimensional structure feature of each immune cell receptor are fused by the state prediction model to obtain a receptor feature of each immune cell receptor, including: For any immune cell receptor in the plurality of immune cell receptors, the gene feature and the sequence feature of the immune cell receptor are spliced by a feature fusioner of the state prediction model to obtain a gene sequence fusion feature of the immune cell receptor. The gene sequence fusion feature and the three-dimensional structure feature of the immune cell receptor are weighted and fused based on a gated attention mechanism to obtain a receptor feature of the immune cell receptor.

20. The apparatus of claim 14, wherein, The state prediction model is used to predict a state based on the receptor feature of each immune cell receptor, and at least one target state of the target object is output, including: A classification module of the state prediction model is used to perform full connection and normalization on the receptor feature of each immune cell receptor, and a probability of each immune cell receptor corresponding to a plurality of candidate states is output. Based on the probability of each immune cell receptor corresponding to the plurality of candidate states, at least one target state of the target object is determined.

21. The apparatus of claim 20, wherein, Based on the probability of each immune cell receptor corresponding to the plurality of candidate states, at least one target state of the target object is determined, including: Based on the probability of each immune cell receptor corresponding to the plurality of candidate states, a reference candidate state corresponding to each immune cell receptor is determined, and the reference candidate state is a candidate state in the plurality of candidate states that meets a first probability condition. Based on the reference candidate state corresponding to the plurality of immune cell receptors, at least one target state of the target object is determined.

22. The apparatus of claim 21, wherein, Based on the reference candidate state corresponding to the plurality of immune cell receptors, at least one target state of the target object is determined, including any one of the following: For a first reference candidate state in the reference candidate state corresponding to the plurality of immune cell receptors, if the number of times that the plurality of immune cell receptors correspond to the first reference candidate state meets a number condition, the first reference candidate state is determined as a target state of the target object. For a second reference candidate state in the reference candidate state corresponding to the plurality of immune cell receptors, an average probability that the plurality of immune cell receptors correspond to the second reference candidate state is determined. If the average probability meets a second probability condition, the second reference candidate state is determined as a target state of the target object.

23. The apparatus of claim 21, wherein, Based on the reference candidate state corresponding to the plurality of immune cell receptors, at least one target state of the target object is determined, including any one of the following: The reference candidate states corresponding to the plurality of immune cell receptors are sorted in order of probability from large to small, and the first target number of reference candidate states are determined as target states of the target object. The reference candidate state corresponding to the plurality of immune cell receptors and the probability corresponding to the reference candidate state are input into a state screening model; the reference candidate state corresponding to the plurality of immune cell receptors and the probability corresponding to the reference candidate state are processed by the state screening model, and the at least one target state is output.

24. The apparatus of claim 14, wherein, Before inputting the gene information, sequence information and three-dimensional structure characteristics of a plurality of immune cell receptors in the immune repertoire of a target object into a state prediction model, the device comprises: For any immune cell receptor in the plurality of immune cell receptors, the amino acid sequence of the CDR3 region of the immune cell receptor is obtained; The amino acid sequence of the CDR3 region of the immune cell receptor is subjected to multiple sequence alignment to obtain at least one reference amino acid sequence, and the similarity between the reference amino acid sequence and the amino acid sequence of the CDR3 region of the immune cell receptor meets a similarity condition; The amino acid sequence of the CDR3 region of the immune cell receptor is obtained, and the homologous template corresponding to the amino acid sequence of the CDR3 region of the immune cell receptor is obtained, wherein the homologous template comprises structural information of the homologous sequence of the amino acid sequence of the CDR3 region of the immune cell receptor; The CDR3 region of the immune cell receptor is obtained, and the homologous template corresponding to the amino acid sequence of the CDR3 region of the immune cell receptor is obtained, wherein the homologous template comprises structural information of the homologous sequence of the amino acid sequence of the CDR3 region of the immune cell receptor; 25. The apparatus of claim 14, wherein, After the state prediction model is used to extract features from the gene information and sequence information of the plurality of immune cell receptors, the device further comprises: The state prediction model is used to fuse the gene features, sequence features, three-dimensional structure features and physicochemical information of amino acids in each immune cell receptor to obtain receptor features of each immune cell receptor.

26. A state prediction model training apparatus characterized by comprising: The device comprises: A three-dimensional structure feature determination unit is configured to, for any immune cell receptor in a plurality of immune cell receptors in an immune repertoire of a sample object, obtain three-dimensional structure information of the immune cell receptor, wherein the three-dimensional structure information comprises three-dimensional coordinates of a plurality of amino acids in the immune cell receptor; The three-dimensional structure feature determination unit is further configured to perform graph convolution on the three-dimensional structure information of the immune cell receptor to obtain three-dimensional structure features of the immune cell receptor, or to encode the three-dimensional structure information of the immune cell receptor based on an attention mechanism to obtain three-dimensional structure features of the immune cell receptor; A training information input unit is configured to input gene information, sequence information and three-dimensional structure characteristics of the plurality of immune cell receptors into a state prediction model; A training feature extraction unit is configured to use the state prediction model to extract features from the gene information and sequence information of the plurality of immune cell receptors to obtain gene features and sequence features of each immune cell receptor; A training feature fusion unit is configured to use the state prediction model to fuse the gene features, sequence features and three-dimensional structure features of each immune cell receptor to obtain receptor features of each immune cell receptor; a sample state output unit, configured to perform state prediction based on the receptor characteristics of each of the immune cell receptors by using the state prediction model, and output at least one sample state of the sample object; a training unit, configured to train the state prediction model based on difference information between the at least one sample state of the sample object and a labeled state.

27. A computer device, comprising: The computer device comprises one or more processors and one or more memories, and the one or more memories store at least one computer program, which is loaded and executed by the one or more processors to implement the state prediction method according to any one of claims 1 to 12, or to implement the training method of the state prediction model according to claim 13.

28. A computer-readable storage medium, characterized in that, The computer readable storage medium stores at least one computer program, which is loaded and executed by the processor to implement the state prediction method according to any one of claims 1 to 12, or to implement the training method of the state prediction model according to claim 13.

29. A computer program product comprising a computer program, characterised in that, The computer program, when executed by the processor, implements the state prediction method according to any one of claims 1 to 12, or implements the training method of the state prediction model according to claim 13.

Images