KR20240076358A - Apparatus and method for generating immunopeptidome pmhc information using artificial intelligence - Google Patents

Abstract

A method of training an artificial intelligence-based prediction model performed by a computing device is disclosed. The method includes obtaining first learning data including first input data corresponding to a peptide and a Major Histocompatibility Complex (MHC) and first labeling data corresponding to the first input data. Step - the first labeling data corresponds to a first number of classes related to the binding of the peptide and the MHC - based on the second learning data to classify a second number of classes related to the binding of the peptide to the MHC Obtaining a first prediction result for the second number of classes from the first learning data using a pre-trained first prediction model, based on the first prediction result and the first labeling data Thus, updating the first labeling data of the first learning data with second labeling data corresponding to a third number of classes related to the combination of peptide and MHC, the second labeling data and the first input data It may include generating third training data including, and training a second prediction model to classify the third number of classes based on the third training data.

Description

Method and device for generating immunopeptidome pMHC information using artificial intelligence technology {APPARATUS AND METHOD FOR GENERATING IMMUNOPEPTIDOME PMHC INFORMATION USING ARTIFICIAL INTELLIGENCE}

The present disclosure relates to artificial intelligence technology, and more specifically, to the generation of information of the peptide-Major Histocompatibility Complex (pMHC), which acts as an immunopeptidome, using artificial intelligence technology. .

The major histocompatibility complex is a locus that encodes ‘MHC molecules’ that function in the immune system. MHC molecules can be of type 1 (class I) and type 2 (class II).

An immunopeptidome refers to a set of peptides expressed on the surface of a cell. For example, an immunopeptidome may refer to a combination of peptides associated with MHC.

Human Leukocyte Antigen (HLA) is a glycoprotein molecule produced by the human Major Histocompatibility Complex (MHC) gene. HLA is not present in mature red blood cells, but is expressed in immature erythroblasts and is expressed on the surface of all tissue cells in the human body, including blood cells such as white blood cells and/or platelets. MHC genes exist in all vertebrates, and the human MHC gene is called an HLA gene, and the product expressed therefrom is called HLA.

MHC genes are involved in recognition of self and non-self, immune response to antigen stimulation, regulation of cellular and humoral immunity, and susceptibility to disease. HLA, a product of the MHC gene, is the second most important antigen after the ABO blood group in the survival of the transplanted organ in solid organ transplantation. HLA is known to play the most important role in the success or failure of bone marrow transplantation. Therefore, immunological recognition of HLA differences can be considered the first step in rejection action against transplanted tissue. Additionally, in transfusion therapy, HLA and antibodies play an important role in the occurrence of various side effects such as platelet transfusion refractoriness, febrile non-hemolytic transfusion side effects, acute lung injury, and post-transfusion graft-versus-host disease.

HLA, like MHC, can be broadly classified into Class I and Class II. Class I is classified into HLA-A, HLA-B, and HLA-C and is expressed in most nucleated cells and platelets. When cytotoxic T cells recognize and eliminate virus-infected cells or tumor cells, they recognize antigens. ) is essential. HLA Class II is classified into HLA-DR, HLA-DQ, and HLA-DP and is expressed in B cells, monocytes, dendritic cells, and activated T cells, and is an antigen receptor for helper T cells. It is known to act as a trigger to induce cellular and humoral immune responses and to be essential when recognizing antigens expressed on antigen-presenting cells. HLA is a gene that shows the greatest polymorphism among genes possessed by humans, and differences in frequency also exist among races and ethnicities.

When a peptide derived from an infectious microorganism-derived protein or a cancer cell-specific protein binds to MHC and is presented on the cell surface, T cells recognize it and trigger an immune response to eliminate the infected cell or cancer cell. In this way, T cells are key regulators (players) that determine specific immune responses to foreign substances that do not exist in the normal human body. Therefore, prediction of TCR (T Cell Receptor) binding to pMHC can be used in the development of personalized vaccines to prevent infectious diseases or cancer.

Republic of Korea registered patent 10-2322832

The present disclosure has been made in response to the above-described background technology, and is intended to predict or identify TCRs capable of binding pMHC in a more efficient manner and/or more accurately.

The technical problems of the present disclosure are not limited to the technical problems mentioned above, and other technical problems not mentioned will be clearly understood by those skilled in the art from the description below.

According to one embodiment of the present disclosure, a method for training an artificial intelligence-based prediction model performed by a computing device is disclosed. The method includes: obtaining first learning data including first input data corresponding to a peptide and a Major Histocompatibility Complex (MHC) and first labeling data corresponding to the first input data. Step - the first labeling data corresponds to a first number of classes related to the binding of the peptide and the MHC - based on the second learning data to classify a second number of classes related to the binding of the peptide to the MHC Obtaining a first prediction result for the second number of classes from the first learning data using a pre-trained first prediction model, based on the first prediction result and the first labeling data Thus, updating the first labeling data of the first learning data with second labeling data corresponding to a third number of classes related to the combination of peptide and MHC, the second labeling data and the first input data It may include generating third training data including, and training a second prediction model to classify the third number of classes based on the third training data.

In one embodiment, the first number of classes includes a first class corresponding to a first score quantitatively indicating the binding possibility between the peptide and the MHC, and a first class corresponding to a second score quantitatively indicating the binding possibility between the peptide and the MHC. It includes two classes, wherein the second number of classes corresponds to a third score that quantitatively represents the binding possibility between the peptide and MHC, and a fourth score that quantitatively represents the binding possibility between the peptide and MHC. a fourth class, and the third score has a value greater than the first score, the first score has a value greater than the second score, and the second score has a value greater than the fourth score. You can have

In one embodiment, at least some of the first number of classes and the second number of classes may be determined based on different criteria.

In one embodiment, the first number of classes may be determined based on data obtained from a first source, and the second number of classes may be determined based on data obtained from a second source. Data obtained from the first source may be less reliable than data obtained from the second source.

In one embodiment, the first learning data includes learning data labeled as binding between a peptide and MHC from data in a public database, and learning data labeled as not binding between a peptide and MHC from data in the public database. It includes, and the second learning data is, learning data labeled as binding between the peptide and MHC from mass spectrometry (MS) data, and learning data labeled as not binding between the peptide and MHC from the MS data. may include.

In one embodiment, the data in the public database may include at least one of the half maximal inhibitory concentration (IC50) of the peptide-MHC complex (pMHC), or the half-life (t _1/2 ) of pMHC binding. .

In one embodiment, the step of updating the first labeling data of the first learning data with the second labeling data corresponding to a third number of classes related to the combination of peptide and MHC includes the first number of classes and updating first labeling data corresponding to the first learning data with the second labeling data corresponding to the third number of classes based on a combination of the second number of classes. . The third number of classes may correspond to the number of possible combinations between the first number of classes and the second number of classes.

In one embodiment, the step of updating the first labeling data of the first learning data with the second labeling data corresponding to the third number of classes related to the combination of peptide and MHC, the first prediction result is If the peptide of the first input data indicates binding to the MHC, and the first labeling data indicates that the peptide of the first input data binds to the MHC, the possibility of binding the first labeling data to the first quantitative value updating with second labeling data indicating that the first prediction result indicates that the peptide of the first input data is bound to the MHC, and the first labeling data indicates that the peptide of the first input data is bound to the MHC. updating the first labeling data with second labeling data indicating a binding possibility of a second quantitative value, if the first prediction result indicates that the peptide and the MHC of the first input data do not bind. Indicates, and if the first labeling data indicates that the peptide of the first input data and MHC bind, updating the first labeling data with second labeling data indicating the binding possibility of a third quantitative value, and If the first prediction result indicates that the peptide of the first input data does not bind to the MHC, and the first labeling data indicates that the peptide of the first input data does not bind to the MHC, the first labeling and updating the data with second labeling data indicating a possible combination of the fourth quantitative value. The first quantitative value has a greater value than the second quantitative value, the second quantitative value has a greater value than the third quantitative value, and the third quantitative value has a greater value than the fourth quantitative value. You can.

In one embodiment, the step of updating the first labeling data of the first learning data with the second labeling data corresponding to a third number of classes related to the combination of peptide and MHC is included in the first prediction result. Based on whether a class corresponds to a class included in the first labeling data, second labeling the first labeling data of the first learning data to correspond to a third number of classes related to the combination of a peptide and MHC. It may include updating with data.

In one embodiment, the step of generating third training data including the second labeling data and the first input data is such that the class included in the first prediction result and the class included in the first labeling data correspond to each other. If so, using the first labeling data as the second labeling data and including the second labeling data and the first input data in the third training data, and a class included in the first prediction result. If the classes included in the first labeling data do not correspond to each other, the method may include not including the first input data in the third learning data.

In one embodiment, the step of generating third learning data including the second labeling data and the first input data is performed when the correct answer data corresponding to the first input data is selected from the first labeling data and the second labeling data. It may include generating the updated third learning data.

In one embodiment, the first prediction model corresponds to an artificial intelligence-based classification model that generates a binary type output for the binding possibility between the peptide and the MHC, and the second prediction model corresponds to the binding possibility between the peptide and the MHC. It can correspond to an artificial intelligence-based classification model that generates three or more types of output.

In one embodiment, the third number may have a value greater than the first number and the third number may have a value greater than the second number.

In one embodiment, the first prediction model or the second prediction model is a Recurrent Neural Network (RNN), a Long Short Term Memory (LSTM) network, a Bidirectional Long Short Term Memory (BiLSTM) network, or a Bidirectional Encoder Representations from Transformers (BERT). ), a Gated Recurrent Unit (GRU), or a Bidirectional Gated Recurrent Unit (BiGRU).

In one embodiment, a tokenization process is applied to the first input data included in the first training data before being input to the first prediction model, and the tokenization process includes amino acid sequences of different lengths. By analyzing the frequency of occurrence for each, tokens having lengths corresponding to each of amino acid sequences of different lengths can be generated. The frequency of appearance is a value that quantitatively represents the probability that a specific amino acid sequence is found within one peptide sequence, or a value that quantitatively represents the ratio of the number of times a specific amino acid sequence is found in all peptides to the number of total peptides. It can be included.

In one embodiment, the tokenization process includes generating a first token group for a first set of amino acid sequences obtained by analyzing the frequency of occurrence for amino acid sequences having a first length, and the first group of tokens for a first set of amino acid sequences. A second set of amino acid sequences obtained by analyzing the frequency of occurrence of amino acid sequences having a second length shorter than the first length, with the first set of amino acid sequences included in the group excluded. It may include the step of generating a second token group including. Tokens included in the first token group may have the first length and tokens included in the second token group may have the second length.

In one embodiment, the tokenization process includes obtaining a list of peptides from an external database or a database included in the computing device, and for each set of amino acids whose amino acid sequence length is K, the first occurrence in the list of peptides Determining a frequency, determining a first set of amino acids whose first occurrence frequency is greater than or equal to a first predetermined threshold among sets of amino acids whose length of the amino acid sequence is K, wherein the length of the amino acid sequence is K-1 determining a second frequency of occurrence within the peptide list for each of the sets of amino acids, wherein, among the sets of amino acids whose amino acid sequences have a length of K-1, the second frequency of occurrence is greater than or equal to a second predetermined threshold; determining a set of two amino acids, and performing tokenization with a length of K for the first set of amino acids and performing tokenization with a length of K-1 for the second set of amino acids. . Here, K is a natural number of 2 or more, and the first threshold value and the second threshold value may have the same or different values.

In one embodiment, the tokenization process includes obtaining a peptide list from an external database or a database included in the computing device, obtaining a CDR3 list from an external database or a database included in the computing device, the length of the amino acid sequence Among amino acid sets where N-M, the M+1 amino acid set whose frequency of occurrence in the peptide list is greater than or equal to a predetermined threshold is included in the token list, and the M+1 amino acid set is removed from the peptide list. , where M is an integer greater than 0, and N-M is a natural number greater than 2, after the step of constructing the token list is performed, the value of M is increased by 1, and determining whether a termination condition is satisfied; if the termination condition is not satisfied, re-performing the step of constructing the token list; and if the termination condition is satisfied, sets of amino acids included in the token list. It may include performing tokenization by generating tokens corresponding to each of the sets of amino acids with corresponding amino acid sequence lengths.

In one embodiment, a computer program stored on a computer-readable storage medium is disclosed. The computer program, when executed by a computing device, causes the computing device to perform operations for learning an artificial intelligence-based prediction model, the operations comprising: first input data corresponding to a peptide and a major histocompatibility complex, and the first input data corresponding to the peptide and the major histocompatibility complex. 1 An operation of acquiring first learning data including first labeling data corresponding to input data, wherein the first labeling data corresponds to a first number of classes related to the binding of the peptide and the MHC, of the peptide and the MHC. Using a first prediction model pre-trained based on the second training data to classify a second number of classes related to the combination, produce a first prediction result for the second number of classes from the first training data An operation of obtaining, based on the first prediction result and the first labeling data, labeling the first labeling data of the first learning data with a second label corresponding to a third number of classes related to the combination of peptide and MHC. updating with data, generating third training data including the second labeling data and the first input data, and classifying the third number of classes based on the third training data. It may include the operation of training a prediction model.

A computing device according to one embodiment is disclosed. The computing device may include at least one processor and memory. The at least one processor is configured to: acquire first learning data including first input data corresponding to a peptide and a major histocompatibility complex and first labeling data corresponding to the first input data - the first labeling data; corresponds to a first number of classes related to the binding of the peptide and the MHC -, using the first prediction model pre-trained based on the second training data to classify the second number of classes related to the binding of the peptide and the MHC Thus, an operation of obtaining a first prediction result for the second number of classes from the first learning data, based on the first prediction result and the first labeling data, a first prediction result of the first learning data Updating labeling data with second labeling data corresponding to a third number of classes related to the combination of a peptide and MHC, generating third learning data including the second labeling data and the first input data. Based on the operation and the third learning data, an operation of training a second prediction model to classify the third number of classes may be performed.

Methods and devices according to an embodiment of the present disclosure can predict or identify TCRs capable of binding pMHC in a more efficient manner and/or more accurately.

1 schematically shows a block diagram of a computing device according to an embodiment of the present disclosure.
2 shows an example structure of an artificial intelligence-based model according to an embodiment of the present disclosure.
Figure 3 exemplarily shows a method of training a prediction model that classifies pMHCs according to the degree of binding according to an embodiment of the present disclosure.
Figure 4 exemplarily shows a method of training a prediction model for binary classifying pMHCs according to the degree of binding according to an embodiment of the present disclosure.
FIG. 5 exemplarily illustrates a method of generating new classes by updating labeling data based on pre-existing labeling data and a prediction result output through a prediction model according to an embodiment of the present disclosure.
6 exemplarily illustrates a method of training a prediction model based on updated labeling data and training data according to an embodiment of the present disclosure.
7 illustratively illustrates a tokenization process according to one embodiment of the present disclosure.
8 is a schematic diagram of a computing environment according to one embodiment of the present disclosure.

Various embodiments are described with reference to the drawings. In this specification, various descriptions are presented to provide an understanding of the disclosure. Before describing specific details for implementing the present disclosure, it should be noted that configurations that are not directly related to the technical gist of the present disclosure have been omitted to the extent that they do not distract from the technical gist of the present invention. In addition, the terms or words used in this specification and claims have meanings that are consistent with the technical idea of the present invention, based on the principle that the inventor can define the concept of appropriate terms in order to explain his or her invention in the best way. It should be interpreted as a concept.

As used herein, the terms "component", "module", "system", "part", etc. refer to a computer-related entity, hardware, firmware, software, a combination of software and hardware, or an implementation of software, and are used interchangeably. It can possibly be used. For example, a component may be, but is not limited to, a process running on a processor, a processor, an object, a thread of execution, a program, and/or a computer. For example, both an application running on a computing device and the computing device can be a component. One or more components may reside within a processor and/or thread of execution. A component may be localized within one computer. A component may be distributed between two or more computers. Additionally, these components can execute from various computer-readable media having various data structures stored thereon. Components may transmit signals, for example, with one or more data packets (e.g., data and/or signals from one component interacting with other components in a local system, a distributed system, to other systems and over a network such as the Internet). Depending on the data being transmitted, they may communicate through local and/or remote processes.

Additionally, the term “or” is intended to mean an inclusive “or” and not an exclusive “or.” That is, unless otherwise specified or clear from context, “X utilizes A or B” is intended to mean one of the natural implicit substitutions. That is, either X uses A; X uses B; Or, if X uses both A and B, “X uses A or B” can apply to either of these cases. Additionally, the term “and/or” as used herein should be understood to refer to and include all possible combinations of one or more of the related listed items.

Additionally, the terms “comprise” and/or “comprising” should be understood to mean that the corresponding feature and/or element is present. However, the terms “comprise” and/or “comprising” should be understood as not excluding the presence or addition of one or more other features, elements and/or groups thereof. Additionally, unless otherwise specified or the context is clear to indicate a singular form, the singular terms herein and in the claims should generally be construed to mean “one or more.”

And, the term “at least one of A or B” or “at least one of A and B” means “if it contains only A,” “if it contains only B,” or “if it is a combination of A and B.” It should be interpreted to mean.

Those skilled in the art will additionally recognize that the various illustrative logical components, blocks, modules, circuits, means, logics, and algorithms described in connection with the embodiments disclosed herein may be implemented using electronic hardware, computer software, or a combination of both. It must be recognized that it can be implemented with To clearly illustrate the interchangeability of hardware and software, various illustrative components, blocks, means, logic, modules, circuits, and steps have been described above generally in terms of their functionality. Whether such functionality is implemented in hardware or software will depend on the specific application and design constraints imposed on the overall system. A skilled technician can implement the described functionality in a variety of ways for each specific application. However, such implementation decisions should not be construed as causing a departure from the scope of the present disclosure.

The description of the presented embodiments is provided to enable anyone skilled in the art to use or practice the present invention. Various modifications to these embodiments will be apparent to those skilled in the art. The general principles defined herein may be applied to other embodiments without departing from the scope of the disclosure. Therefore, the present invention is not limited to the embodiments presented herein. The present invention is to be interpreted in the broadest scope consistent with the principles and novel features presented herein.

In the present disclosure, terms represented by N, such as first, second, or third, are used to distinguish at least one entity. For example, the entities expressed as first and second may be the same or different from each other.

In this disclosure, for convenience of explanation, human leukocyte antigen (HLA) is used as an example of MHC. Therefore, the description of HLA or MHC used below is an example for expressing the description of MHC or HLA, and the scope of rights of the present disclosure will be determined based on the content stated in the claims, and through examples of HLA The scope of the rights should not be interpreted as limited to HLA. As such, HLA and MHC in the present disclosure may be used interchangeably.

The term “human leukocyte antigen (HLA)” used in this disclosure is a glycoprotein molecule produced by the human MHC gene, and is a gene that shows the greatest polymorphism among genes possessed by humans. HLA typing, which determines HLA type, can be actively used in various fields such as organ transplantation, immunotherapy, disease-related research, paternity tests such as paternity determination, forensic use, and genetic research.

HLA types in the present disclosure may include, for example, HLA-A type, HLA-B type, and/or HLA-C type.

In one embodiment, the complex of MHC and peptide may refer to a peptide antigen presentation complex through MHC class I molecules after processing through the proteasome in an antigen presenting cell (APC). The proteasome is composed of two monomers, LMP-2 and LMP-7 (low molecular weight polypeptide). These two proteasome monomers are located near the TAP-1 and TAP-2 genes within the MHC gene. Proteasome monomers are particularly important for the degradation of peptides that bind to MHC I molecules. When cells are treated with the cytokine interferon gamma (IFN-γ), the expression of LMP-2 and LMP-7 can be induced. Expression of LMP-2 and LMP-7 changes the substrate specificity of the proteasome, increasing its ability to decompose into peptides. The expression of proteins related to antigen presentation, such as MHC I and MECL-1, as well as LMP proteins, is increased by IFN-γ, which can increase antigen presentation in antigen-presenting cells.

MHC I molecules bind not only to peptide antigens degraded by the proteasome, but also to peptides produced by proteolytic enzymes present in the endoplasmic reticulum.

Tapasin acts as a bridge between TAP-1 (transporter associated with antigen processing) and MHC I, which has a stable tertiary structure within the endoplasmic reticulum (ER), and when a peptide enters, the MHC I complex is bound to the tapasin. It leaves the Shin and TAP proteins and becomes a complete peptide-MHC class I complex.

T cells only recognize external antigens presented on the cell surface, and the activation of mature peripheral T cells begins through the interaction between the antigen and peptide presented in the peptide binding niche of the TCR and MHC molecules.

1 schematically shows a block diagram of a computing device 100 according to an embodiment of the present disclosure.

Computing device 100 according to an embodiment of the present disclosure may include a processor 110 and a memory 130.

The configuration of the computing device 100 shown in FIG. 1 is only a simplified example. In one embodiment of the present disclosure, the computing device 100 may include different components for performing the computing environment of the computing device 100, and only some of the disclosed components may configure the computing device 100.

The computing device 100 in the present disclosure may refer to any type of node constituting a system for implementing embodiments of the present disclosure. Computing device 100 may refer to any type of user terminal or any type of server. The components of the computing device 100 described above are exemplary and some may be excluded or additional components may be included. For example, when the above-described computing device 100 includes a user terminal, an output unit (not shown) and an input unit (not shown) may be included within the scope of the computing device 100.

The computing device 100 in the present disclosure may perform technical features according to embodiments of the present disclosure, which will be described later. For example, the computing device 100 uses an artificial intelligence-based prediction model that uses input data corresponding to MHC and peptides to generate a prediction result including the degree of binding of MHC and peptides corresponding to the input data. . For example, the computing device 100 obtains information about MHC and peptides from a sample obtained from a subject, and stores information on the CDR3 of the TCR based on mass spectrometry (MS) data for the peptide-MHC conjugate. Corresponding prediction results can be generated.

In one embodiment of the present disclosure, the computing device 100 may obtain a result of performing base sequence analysis (eg, Next Generation Sequencing) from a server or an external entity. In another embodiment, the computing device 100 may perform base sequence analysis on genetic data (eg, DNA or RNA) obtained from a biological sample derived from a subject. As the term used in the present disclosure, base sequencing may be performed by any type of technique capable of analyzing the sequence of bases, for example, whole genome sequencing, whole exome It may include, but is not limited to, whole exome sequencing or whole transcriptome sequencing.

As used in the present disclosure, the term "subject" may refer to a subject or individual for obtaining a biological sample containing a major histocompatibility complex (MHC), a peptide, and/or a complex thereof.

The terms and samples used in the present disclosure can be used without limitation as long as they are obtained from an individual or subject whose MHC type is to be determined, for example, cells or tissues, blood, whole blood, serum, plasma obtained by biopsy, etc. , saliva, cerebrospinal fluid, various secretions, urine and/or feces, etc. Preferably, the sample may be selected from the group consisting of blood, plasma, serum, saliva, nasal fluid, sputum, ascites, vaginal secretions and/or urine, and more preferably may be blood, plasma or serum. The sample may be pretreated prior to use for detection or diagnosis. For example, pretreatment methods may include homogenization, filtration, distillation, extraction, concentration, inactivation of interfering components, and/or addition of reagents, etc. In the present disclosure, the biological sample may be, but is not limited to, tissue, cells, whole blood, and/or blood.

In one embodiment, the processor 110 may consist of at least one core, including a central processing unit (CPU) and a general purpose graphics processing unit (GPGPU) of the computing device 100. , may include a processor for data analysis and/or processing, such as a tensor processing unit (TPU).

The processor 110 may read the computer program stored in the memory 130 and generate a prediction result including the degree of pMHC binding, according to an embodiment of the present disclosure.

According to an embodiment of the present disclosure, the processor 110 may perform an operation for learning a neural network. The processor 110 is used for learning neural networks, such as processing input data for learning in deep learning (DL), extracting features from input data, calculating errors, and updating the weights of the neural network using backpropagation. Calculations can be performed. At least one of the CPU, GPGPU, and TPU of the processor 110 may process learning of the network function. For example, CPU and GPGPU can work together to process learning of network functions and data classification using network functions. Additionally, in one embodiment of the present disclosure, processors of a plurality of computing devices may be used together to process learning of a network function and data classification using a network function. Additionally, a computer program executed in a computing device according to an embodiment of the present disclosure may be a CPU, GPGPU, or TPU executable program.

Additionally, processor 110 may typically handle overall operations of computing device 100. For example, the processor 110 processes data, information, or signals input or output through components included in the computing device 100 or runs an application program stored in the storage to provide information or information appropriate to the user. Functions can be provided or processed.

According to one embodiment of the present disclosure, the memory 130 may store any type of information generated or determined by the processor 110 and any type of information received by the computing device 100. According to an embodiment of the present disclosure, the memory 130 may be a storage medium that stores computer software that allows the processor 110 to perform operations according to embodiments of the present disclosure. Accordingly, the memory 130 may refer to computer readable media for storing software codes required to perform embodiments of the present disclosure, data to be executed by the codes, and execution results of the codes.

According to one embodiment of the present disclosure, the memory 130 may refer to any type of storage medium. For example, the memory 130 may be a flash memory type or a hard disk type. ), multimedia card micro type, card type memory (e.g. SD or XD memory, etc.), RAM (Random Access Memory), SRAM (Static Random Access Memory), ROM (Read-Only) Memory, ROM), EEPROM (Electrically Erasable Programmable Read-Only Memory), PROM (Programmable Read-Only Memory), magnetic memory, magnetic disk, and optical disk. The computing device 100 may operate in connection with web storage that performs a storage function of the memory 130 on the Internet. The description of the memory described above is only an example, and the memory 130 used in the present disclosure is not limited to the examples described above.

The communication unit (not shown) in the present disclosure can be configured regardless of the communication mode, such as wired or wireless, and can be used in various communication networks such as a personal area network (PAN) and a wide area network (WAN). It can be configured. In addition, the network unit 150 can operate based on the well-known World Wide Web (WWW), and is a wireless transmission technology used for short-distance communication such as Infrared Data Association (IrDA) or Bluetooth. You can also use .

Computing device 100 in the present disclosure may include any type of user terminal and/or any type of server. Accordingly, embodiments of the present disclosure may be performed by a server and/or a user terminal.

A user terminal may include any type of terminal capable of interacting with a server or other computing device. User terminals include, for example, mobile phones, smart phones, laptop computers, personal digital assistants (PDAs), slate PCs, tablet PCs, and ultrabooks. It can be included.

Servers may include any type of computing system or computing device, such as, for example, microprocessors, mainframe computers, digital processors, portable devices, and device controllers.

In an additional embodiment, the above-described server may mean an entity that stores and manages immunopeptidome information, peptide sequence information, base sequence information, or genetic information. The server stores immunopeptidome information, peptide sequence information, amino acid identifier information for each position, nucleotide sequence information, genetic information, or reliability information of databases (e.g., IEDB, HLA atlas Ligand, SysteMHC Altas). It may include a storage unit (not shown), and the storage unit may be included in the server or may exist under the management of the server. As another example, the storage unit may be implemented in a form that exists outside the server and can communicate with the server. In this case, the storage may be managed and controlled by an external server that is different from the server.

2 shows an example structure of an artificial intelligence-based model according to an embodiment of the present disclosure.

Throughout this specification, prediction model, artificial intelligence-based prediction model, artificial intelligence model, artificial intelligence-based model, computational model, neural network, network function, and neural network may be used with the same meaning.

A neural network can generally consist of a set of interconnected computational units, which can be referred to as nodes. These nodes may also be referred to as neurons. A neural network consists of at least one node. Nodes (or neurons) that make up neural networks may be interconnected by one or more links.

Within a neural network, one or more nodes connected through a link may form a relative input node and output node relationship. The concepts of input node and output node are relative, and any node in an output node relationship with one node may be in an input node relationship with another node, and vice versa. As described above, input node to output node relationships can be created around links. One or more output nodes can be connected to one input node through a link, and vice versa.

In a relationship between an input node and an output node connected through one link, the value of the data of the output node may be determined based on the data input to the input node. Here, the link connecting the input node and the output node may have a weight. Weights may be variable and may be varied by the user or algorithm in order for the neural network to perform the desired function. For example, when one or more input nodes are connected to one output node by respective links, the output node is set to the values input to the input nodes connected to the output node and the links corresponding to each input node. The output node value can be determined based on the weight.

As described above, in a neural network, one or more nodes are interconnected through one or more links to form an input node and output node relationship within the neural network. The characteristics of the neural network may be determined according to the number of nodes and links within the neural network, the correlation between the nodes and links, and the value of the weight assigned to each link. For example, if the same number of nodes and links exist and two neural networks with different weight values of the links exist, the two neural networks may be recognized as different from each other.

A neural network may consist of a set of one or more nodes. A subset of nodes that make up a neural network can form a layer. Some of the nodes constituting the neural network may form one layer based on the distances from the first input node. For example, a set of nodes with a distance n from the initial input node may constitute n layers. The distance from the initial input node can be defined by the minimum number of links that must be passed to reach the node from the initial input node. However, this definition of a layer is arbitrary for explanation purposes, and the order of a layer within a neural network may be defined in a different way than described above. For example, a layer of nodes may be defined by distance from the final output node.

In one embodiment of the present disclosure, a set of neurons or nodes may be defined by the expression layer.

The initial input node may refer to one or more nodes in the neural network through which data is directly input without going through links in relationships with other nodes. Alternatively, in the relationship between nodes based on links within a neural network, it may refer to nodes that do not have other input nodes connected by links. Similarly, the final output node may refer to one or more nodes that do not have an output node in their relationship with other nodes among the nodes in the neural network. Additionally, hidden nodes may refer to nodes constituting a neural network other than the first input node and the last output node.

The neural network according to an embodiment of the present disclosure is a neural network in which the number of nodes in the input layer may be the same as the number of nodes in the output layer, and the number of nodes decreases and then increases again as it progresses from the input layer to the hidden layer. You can. In addition, the neural network according to another embodiment of the present disclosure may be a neural network in which the number of nodes in the input layer may be less than the number of nodes in the output layer, and the number of nodes decreases as it progresses from the input layer to the hidden layer. there is. In addition, the neural network according to another embodiment of the present disclosure may be a neural network in which the number of nodes in the input layer may be greater than the number of nodes in the output layer, and the number of nodes increases as it progresses from the input layer to the hidden layer. You can. A neural network according to another embodiment of the present disclosure may be a neural network that is a combination of the above-described neural networks.

A deep neural network (DNN) may refer to a neural network that includes multiple hidden layers in addition to the input layer and output layer. Deep neural networks allow you to identify latent structures in data. That is, the potential structure of a photo, text, video, voice, protein sequence structure, gene sequence structure, peptide sequence structure, music (e.g., which object is in the photo, what is the content and emotion of the text, the voice (what the content and emotions are, etc.), and/or the binding affinity between the peptide and MHC. Deep neural networks include convolutional neural networks (CNNs), recurrent neural networks (RNNs), auto encoders, generative adversarial networks (GANs), and restricted Boltzmann machines ( It may include restricted boltzmann machine (RBM), deep belief network (DBN), Q network, U network, Siamese network, etc. The description of the deep neural network described above is only an example and the present disclosure is not limited thereto.

For example, the artificial intelligence-based prediction model of the present disclosure includes Recurrent Neural Network (RNN), Long Short Term Memory (LSTM) network, Bidirectional Long Short Term Memory (BiLSTM) network, Bidirectional Encoder Representations from Transformers (BERT), It may include a Gated Recurrent Unit (GRU) or a Bidirectional Gated Recurrent Unit (BiGRU).

The artificial intelligence-based prediction model of the present disclosure can be expressed by a network structure of any of the structures described above, including an input layer, a hidden layer, and an output layer.

Neural networks that can be used in the artificial intelligence-based model of the present disclosure include supervised learning, unsupervised learning, semi-supervised learning, transfer learning, or reinforcement learning. It can be learned in at least one of the following ways: Learning of a neural network may be a process of applying knowledge for the neural network to perform a specific operation to the neural network. For example, a prediction model according to an embodiment of the present disclosure applies a mask to some amino acids among the amino acid sequences and then uses semi-supervised learning to match the masked amino acids. It can be learned in this way.

Neural networks can be trained to minimize output errors. In neural network learning, learning data is repeatedly input into the neural network, the output of the neural network and the error of the target for the learning data are calculated, and the error of the neural network is transferred from the output layer of the neural network to the input layer in the direction of reducing the error. This is the process of updating the weight of each node in the neural network through backpropagation. In the case of supervised learning, learning data in which the correct answer is labeled for each learning data is used (i.e., labeled learning data), and in the case of unsupervised learning, the correct answer may not be labeled in each learning data. That is, for example, in the case of supervised learning on data classification, the learning data may be data in which each training data is labeled with a category. Labeled training data is input to the neural network, and the error can be calculated by comparing the output (category) of the neural network with the label of the training data. As another example, in the case of unsupervised learning on data classification, the error can be calculated by comparing the input training data with the neural network output. The calculated error is backpropagated in the reverse direction (i.e., from the output layer to the input layer) in the neural network, and the connection weight of each node in each layer of the neural network can be updated according to backpropagation. The amount of change in the connection weight of each updated node may be determined according to the learning rate. The neural network's calculation of input data and backpropagation of errors can constitute a learning cycle (epoch). The learning rate may be applied differently depending on the number of repetitions of the learning cycle of the neural network. For example, in the early stages of neural network training, a high learning rate can be used to increase efficiency by allowing the neural network to quickly achieve a certain level of performance, and in the later stages of training, a low learning rate can be used to increase accuracy.

In the learning of neural networks, the training data can generally be a subset of real data (i.e., the data to be processed using the learned neural network), and thus the error for the training data is reduced, but the error for the real data is reduced. There may be an incremental learning cycle. Overfitting is a phenomenon in which errors in actual data increase due to excessive learning on training data. For example, a phenomenon in which a neural network that learned a cat by showing a yellow cat fails to recognize that it is a cat when it sees a non-yellow cat may be a type of overfitting. Overfitting can cause errors in machine learning algorithms to increase. To prevent such overfitting, various optimization methods can be used. To prevent overfitting, methods such as increasing the training data, regularization, dropout to disable some of the network nodes during the learning process, and use of a batch normalization layer are used. It can be applied.

A computer-readable medium storing a data structure according to an embodiment of the present disclosure is disclosed. The above-described data structure can be stored in a storage unit in the present disclosure, executed by a processor, and transmitted and received by a communication unit.

Data structure can refer to the organization, management, and storage of data to enable efficient access and modification of data. Data structure can refer to the organization of data to solve a specific problem (e.g., retrieving data, storing data, or modifying data in the shortest possible time). A data structure may be defined as a physical or logical relationship between data elements designed to support a specific data processing function. Logical relationships between data elements may include connection relationships between user-defined data elements. Physical relationships between data elements may include actual relationships between data elements that are physically stored in a computer-readable storage medium (e.g., a persistent storage device). A data structure may specifically include a set of data, relationships between data, and functions or instructions applicable to the data. Effectively designed data structures allow computing devices to perform computations while minimizing the use of the computing device's resources. Specifically, computing devices can increase the efficiency of operations, reading, insertion, deletion, comparison, exchange, and search through effectively designed data structures.

Data structures can be divided into linear data structures and non-linear data structures depending on the type of data structure. A linear data structure may be a structure in which only one piece of data is connected to another piece of data. Linear data structures may include List, Stack, Queue, and Deque. A list can refer to a set of data that has an internal order. The list may include a linked list. A linked list may be a data structure in which data is connected in such a way that each data is connected in a single line with a pointer. In a linked list, a pointer may contain connection information to the next or previous data. Depending on its form, a linked list can be expressed as a singly linked list, a doubly linked list, or a circularly linked list. A stack may be a data listing structure that allows limited access to data. A stack can be a linear data structure in which data can be processed (for example, inserted or deleted) at only one end of the data structure. Data stored in the stack may have a data structure (LIFO-Last in First Out) where the later it enters, the sooner it comes out. A queue is a data listing structure that allows limited access to data. Unlike the stack, it can be a data structure (FIFO-First in First Out) where data stored later is released later. A deck can be a data structure that can process data at both ends of the data structure.

A non-linear data structure may be a structure in which multiple pieces of data are connected behind one piece of data. Nonlinear data structures may include graph data structures. A graph data structure can be defined by vertices and edges, and an edge can include a line connecting two different vertices. Graph data structure may include a tree data structure. A tree data structure may be a data structure in which there is only one path connecting two different vertices among a plurality of vertices included in the tree. In other words, it may be a data structure that does not form a loop in the graph data structure.

Throughout this specification, artificial intelligence-based model, computational model, neural network, network function, and neural network may be used with the same meaning. Below, it is described in a unified manner as a neural network. Data structures may include neural networks. And the data structure including the neural network may be stored in a computer-readable medium. Data structures including neural networks also include data preprocessed for processing by a neural network, data input to the neural network, weights of the neural network, hyperparameters of the neural network, data acquired from the neural network, activation functions associated with each node or layer of the neural network, neural network It may include a loss function for learning. A data structure containing a neural network may include any of the components disclosed above. In other words, the data structure including the neural network includes preprocessed data for processing by the neural network, data input to the neural network, weights of the neural network, hyperparameters of the neural network, data acquired from the neural network, activation functions associated with each node or layer of the neural network, neural network It may be configured to include all or any combination of the loss function for learning. In addition to the configurations described above, a data structure containing a neural network may include any other information that determines the characteristics of the neural network. Additionally, the data structure may include all types of data used or generated in the computational process of a neural network and is not limited to the above. Computer-readable media may include computer-readable recording media and/or computer-readable transmission media. A neural network can generally consist of a set of interconnected computational units, which can be referred to as nodes. These nodes may also be referred to as neurons. A neural network consists of at least one node.

The data structure may include data input to the neural network. A data structure containing data input to a neural network may be stored in a computer-readable medium. Data input to the neural network may include learning data input during the neural network learning process and/or input data input to the neural network on which training has been completed. Data input to the neural network may include data that has undergone pre-processing and/or data subject to pre-processing. Preprocessing may include a data processing process to input data into a neural network. Therefore, the data structure may include data subject to preprocessing and data generated by preprocessing. The above-described data structure is only an example and the present disclosure is not limited thereto.

The data structure may include the weights of the neural network. (In this specification, weights and parameters may be used with the same meaning.) And the data structure including the weights of the neural network may be stored in a computer-readable medium. A neural network may include multiple weights. Weights may be variable and may be varied by the user or algorithm in order for the neural network to perform the desired function. For example, when one or more input nodes are connected to one output node by respective links, the output node is set to the values input to the input nodes connected to the output node and the links corresponding to each input node. Based on the weight, the data value output from the output node can be determined. The above-described data structure is only an example and the present disclosure is not limited thereto.

As an example and not a limitation, the weights may include weights that are changed during the neural network learning process and/or weights for which neural network learning has been completed. Weights that change during the neural network learning process may include weights that change at the start of the learning cycle and/or weights that change during the learning cycle. Weights for which neural network training has been completed may include weights for which a learning cycle has been completed. Therefore, the data structure including the weights of the neural network may include weights that are changed during the neural network learning process and/or the data structure including the weights for which neural network learning has been completed. Therefore, the above-mentioned weights and/or combinations of each weight are included in the data structure including the weights of the neural network. The above-described data structure is only an example and the present disclosure is not limited thereto.

The data structure including the weights of the neural network may be stored in a computer-readable storage medium (e.g., memory, hard disk) after going through a serialization process. Serialization can be the process of converting a data structure into a form that can be stored on the same or a different computing device and later reorganized and used. Computing devices can transmit and receive data over a network by serializing data structures. Data structures containing the weights of a serialized neural network can be reconstructed on the same computing device or on a different computing device through deserialization. The data structure including the weights of the neural network is not limited to serialization. Furthermore, the data structure including the weights of the neural network is a data structure to increase computational efficiency while minimizing the use of computing device resources (e.g., B-Tree, R-Tree, Trie, m-way search tree in non-linear data structures). , AVL tree, Red-Black Tree). The foregoing is merely an example and the present disclosure is not limited thereto.

The data structure may include hyper-parameters of a neural network. And the data structure including the hyperparameters of the neural network can be stored in a computer-readable medium. A hyperparameter may be a variable that can be changed by the user. Hyperparameters include, for example, learning rate, cost function, number of learning cycle repetitions, weight initialization (e.g., setting the range of weight values subject to weight initialization), Hidden Unit. It may include a number (e.g., number of hidden layers, number of nodes in hidden layers). The above-described data structure is only an example and the present disclosure is not limited thereto.

A transformer may be considered as a network function for the prediction model according to an embodiment of the present disclosure. As an example, the prediction model may operate based on a transformer. This prediction model may be operated using, for example, a recurrent neural network to which an attention algorithm is applied or a transformer to which an attention algorithm is applied.

In one embodiment, the transformer may be comprised of an encoder that encodes the embedded data and a decoder that decodes the encoded data. A transformer may have a structure that receives a series of data, goes through encoding and decoding steps, and outputs a series of data of different types. In one embodiment, a series of data can be processed into a form that can be operated by a transformer. The process of processing a series of data into a form that can be operated by a transformer may include an embedding process. Expressions such as data token, embedding vector, embedding token, etc. may refer to data embedded in a form that a transformer can process.

In order for the transformer to encode and decode a series of data, the encoders and decoders within the transformer can be processed using an attention algorithm. The attention algorithm calculates the similarity of one or more keys for a given query, reflects the given similarity to the value corresponding to each key, and returns the reflected similarity value ( It may refer to an algorithm that calculates the attention value by weighting the values.

Depending on how the query, key, and value are set, various types of attention algorithms can be classified. For example, if attention is obtained by setting the query, key, and value all the same, this may mean a self-attention algorithm. In order to process a series of input data in parallel, when the dimension of the embedding vector is reduced and attention is obtained by obtaining individual attention heads for each divided embedding vector, this refers to a multi-head attention algorithm. can do.

In one embodiment, the transformer may be composed of modules that perform a plurality of multi-head self-attention algorithms or multi-head encoder-decoder algorithms. In one embodiment, the transformer may also include additional components other than the attention algorithm, such as an embedding layer, a normalization layer, and a softmax layer. A method of configuring a transformer using an attention algorithm may include the method disclosed in Vaswani et al., Attention Is All You Need, 2017 NIPS, which is incorporated herein by reference.

Transformers can be applied to various data domains such as embedded natural language, embedded sequence information, segmented image data, and audio waveforms to convert a series of input data into a series of output data. In order to convert data with various data domains into a series of data that can be input to the transformer, the transformer can embed the data. Transformers can process additional data expressing relative positional or phase relationships between a series of input data. Alternatively, the series of input data may be embedded by additionally reflecting vectors expressing relative positional relationships or phase relationships between the input data. In one example, the relative positional relationship between a series of input data may include, but is not limited to, word order within a natural language sentence, relative positional relationship of each segmented image, temporal order of segmented audio waveforms, etc. . The process of adding information expressing relative positional or phase relationships between a series of input data may be referred to as positional encoding.

An example of a method for embedding data and converting it to a transformer is disclosed in Dosovitskiy, et al., AN IMAGE IS WORTH 16X16 WORDS: TRANSFORMERS FOR IMAGE RECOGNITION AT SCALE, which document is incorporated herein by reference.

Figure 3 exemplarily shows a method of training a prediction model that classifies pMHCs according to the degree of binding according to an embodiment of the present disclosure.

In one embodiment, the steps shown in FIG. 3 may be performed by computing device 100. In a further embodiment, the steps shown in FIG. 3 may be implemented by multiple entities, such that some of the steps shown in FIG. 3 are performed at a user terminal and others are performed at a server.

In one embodiment, the computing device 100 acquires first learning data including first input data corresponding to a peptide-major histocompatibility complex conjugate (pMHC) and first labeling data corresponding to the first input data. You can do it (S310).

For example, the first input data may include sequences of peptides corresponding to the pMHC and amino acids of the MHC. As another example, the amino acid sequences of the peptide and MHC may be integrated into each other. Accordingly, the integrated input data may represent an amino acid sequence representing pMHC.

In one example, an amino acid sequence may include a group of identifiers that represent amino acids, for example.

As another example, the MHC data of the first input data may include the form of a nucleotide sequence, a type of human leukocyte antigen (HLA), or a DNA (DeoxyriboNucleic Acid) sequence encoding HLA. there is.

In one embodiment, the first input data may be obtained from public databases and/or experimental results data. For example, the experimental result data of the first input data is MHC MS analysis method (Mass Spectrometry), LC-MS/MS method (Liquid Chromatography-MS/MS analysis), Sanger sequencing method, and Edman Degradation method. ), or PMF analysis (Peptide Mass Fingerprinting), etc., but is not limited to this.

In a further embodiment, the input data may include input data to which Blosum encoding or one-hot encoding has been applied to the amino acid sequence.

In a further embodiment, the input data may be a factor indicating the polarity of amino acids, a factor indicating the molecular weight of the amino acids, a factor indicating whether the amino acids are hydrophobic, hydrophilic, or neutral, a factor regarding the properties of electrical charge of the amino acids, or an amino acid. It may additionally include at least one factor among factors indicating whether they are aromatic or aliphatic.

In one embodiment, the computing device 100 may acquire first learning data including first labeling data corresponding to first input data. Through the labeling process, pMHCs included in the first input data can be labeled into a plurality of classes based on the degree of binding between the peptide and MHC. In one embodiment, a plurality of classes may correspond to a first number of classes. At least some of the first number of classes may be different from at least some of the second number of classes used in the pre-training process of the first prediction model.

For example, the labeling data may have six types of labeling, where the 1a label has a peptide-MHC binding degree of 100%, the 1b label has a peptide-MHC binding degree of 80%, and the 1c label has a peptide-MHC binding degree of 100%. The MHC binding degree is 60%, the 1d label may indicate a peptide-MHC binding degree of 40%, the 1e label may indicate a peptide-MHC binding degree of 20%, and the 1f label may indicate a peptide-MHC binding degree of 0%.

In an additional embodiment, the process of acquiring the first labeling data is such that the pMHCs included in the first input data are labeled with a degree of binding between the peptide and MHC of 100% when the IC50 (half maximum inhibitory concentration) is 500 nM or less, and the IC50 is 500 nM or less. If it exceeds 500nM, the degree of binding between the peptide and MHC can be labeled as 0%.

As an additional example, pMHCs included in the first input data are labeled with a binding degree of 100% between the peptide and MHC when the IC50 is 250 nM or less, and when the IC50 is greater than 250 nM and less than 500 nM, the binding degree between the peptide and MHC is labeled as 67%. If it is above 500nM and below 750nM, the degree of binding between the peptide and MHC may be labeled as 33%, and if it is above 750nM, the degree of binding between the peptide and MHC may be labeled as 0%. The indicator indicating the degree of pMHC binding is not limited to the IC50.

In one embodiment, the computing device 100 may generate first learning data by performing preprocessing including a tokenization process and a segmenting process on the first input data. .

Additional description of the tokenization process and segmenting process will be described later in FIG. 7.

In one embodiment, computing device 100 uses a first prediction model pre-trained based on the second training data to classify a second number of classes associated with the binding of a peptide and an MHC, from the first training data, The first prediction result for the second number of classes may be obtained (S320).

In one embodiment, a first prediction model pre-trained based on the second training data to classify a second number of classes related to peptide-MHC binding may be used to obtain the first prediction result.

In one embodiment, the first labeling data may correspond to a first number of classes associated with binding of a peptide to MHC. The first number of classes may include a first class corresponding to a first score quantitatively indicating the binding possibility between the peptide and MHC, and a second class corresponding to a second score quantitatively indicating the binding possibility between the peptide and MHC. You can. In a further example, the second number of classes is a third class corresponding to a third score quantitatively representing the binding potential between the peptide and the MHC, and a fourth class corresponding to a fourth score quantitatively representing the binding probability between the peptide and the MHC. may include.

In one embodiment, the first training data to be input to the first prediction model may be generated based on the first labeling data, and the first prediction model may be generated based on the second training data based on the second labeling data in the pre-training process. So it can be learned in advance. In this way, the technique according to an embodiment of the present disclosure pre-trains a first prediction model according to a specific labeling standard, and input data or training data labeled according to a standard different from the specific labeling standard is used as a first prediction model for which pre-training has been completed. The first prediction result can be obtained by inputting it into the prediction model. Here, the first prediction result may include a second number of classes corresponding to second training data on which the first prediction model was pre-trained.

A description of the scores that quantitatively represent the number of classes and the possibility of binding between peptides and MHC will be described later in FIGS. 4, 5, and 6.

In one embodiment, the second learning data is training data labeled as binding between a peptide and MHC from MS data (e.g., LC-MS/MS data) and training data labeled as not binding between a peptide and MHC from the MS data. may include training data. In an additional embodiment, the second learning data may be determined by a separate artificial intelligence-based model that uses peptide and MHC as input and the degree of binding as output.

In one embodiment, the reliability of the data used to construct the first learning data and the reliability of the data used to construct the second learning data may be different from each other. For example, the second training data used to pre-train the first prediction model may be labeled based on data with relatively high reliability, and the first training data input to the first prediction model for which pre-training has been completed Labeling may be done based on relatively unreliable data.

In one embodiment, the first prediction model may be pre-trained to classify a second number (eg, two) of classes based on second training data. In a further embodiment, the pre-training of the first prediction model includes the correct answer of pMHC (Binding pMHC) identified as having occurred through experimental data and pMHC (Non-binding pMHC) identified as having not occurred through experimental data. Data sets are available.

Pre-learning of the first prediction model will be described later in FIG. 4.

In one embodiment, the computing device 100 may obtain a first prediction result for a second number of classes from first training data using a pre-trained first prediction model.

In one embodiment, the first training data is training data labeled as binding between a peptide and an MHC from data in a public database (e.g., IEDB, HLA atlas Ligand, SysteMHC, Altas), and training data labeled as binding a peptide and an MHC from data in a public database (e.g., IEDB, HLA atlas Ligand, SysteMHC, Altas). may contain training data labeled as non-binding. In an additional embodiment, the first learning data may be determined by a separate artificial intelligence-based model that uses peptide and MHC as input and the degree of binding as output.

In one embodiment, the first prediction result output from the first learning data using the first prediction model may refine the degree of binding from data including the degree of pMHC binding in a public database. Data containing a fine-grained degree of combination can be classified into new classes. For example, the first prediction result output using the first prediction model can be viewed as data cleansing of the second learning data by the first prediction model. Data purification may be a task to increase the reliability of data on the degree of pMHC binding by processing values for the degree of pMHC binding (e.g., IC50, t _1/2 ).

In one embodiment, the computing device 100, based on the first prediction result and the first labeling data, classifies the first labeling data of the first learning data to correspond to a third number of classes related to the combination of the peptide and the MHC. It can be updated with the second labeling data (S330).

In one embodiment, updating labeling data may be used to encompass changing the labeling standard of the labeling data corresponding to the first input data, changing the labeling method, or changing the labeling value.

In one embodiment, the first prediction result may include the results of pMHCs classified based on information including the degree of binding of pMHC obtained through the first prediction model.

In one embodiment, the first labeling data may correspond to first input data corresponding to peptides and MHC. For example, the first labeling data may be 1a labeling indicating 100% binding, 1b labeling indicating 80% binding, and 60% binding depending on the degree of binding of pMHCs of the first input data included in the first learning data. A 1c labeling, a 1d labeling indicating 40% binding, a 1e labeling indicating 20% binding, and a 1f labeling indicating 0% binding.

In one embodiment, based on a combination of the first number of classes and the second number of classes, first labeling data corresponding to the first training data is converted into second labeling data corresponding to the third number of classes. It can be updated.

In one embodiment, the number in the third number of classes may mean a natural number of 3 or more. In a further embodiment, the second labeling data corresponding to a third number (e.g., 4) of classes is class 1', e.g., class 1', meaning quantified binding of pMHC, quantified weak binding of pMHC. It may correspond to class 2', meaning quantified weak non-binding of pMHC, class 3', meaning quantified weak non-binding of pMHC, and class 4', meaning quantified non-binding of pMHC. In this example, a first number of classes may include two classes representing association or disunion obtained according to a low-confidence test, and a second number of classes may include two classes representing association or non-association obtained according to a high-confidence test. It can contain two classes representing a combination. The output of the first prediction model pretrained with the second number of classes will include the second number of classes, and the first number of classes labeled in the input first training data with this second number of classes. By comparing, new types of classes to be labeled in the first input data can be created. In this example, a total of four classes can be created as third classes according to a combination of two classes corresponding to the first classes and two classes corresponding to the second classes. Accordingly, first input data to which labeling based on four classes is applied may constitute third learning data, and the first prediction model is retrained or the second prediction model is retrained based on this third learning data. It can be trained. In this way, the accuracy of labeling results according to the results of tests with relatively low accuracy or reliability can be increased, and the technical effect that a sufficient amount of learning data can be acquired and generated can be achieved.

Additional description of the step of updating the first labeling data to the second labeling data will be described later with reference to FIG. 5.

In one embodiment, the computing device 100 may generate third learning data including second labeling data and first input data (S340).

In one embodiment, the third training data may be used to train a second prediction model to classify the third number of classes. In a further embodiment, the third training data may be used to retrain the first prediction model to classify a third number of classes. Accordingly, the first prediction model and/or the second prediction model can be learned to be more accurately segmented into a third number of classes, so that the inference performance of the first prediction model and/or the second prediction model as a result of such learning This can be improved.

In a further embodiment, the step of generating third training data including second labeling data and first input data may be performed when the class included in the first prediction result and the class included in the first labeling data correspond to each other. , the first labeling data may be used as the second labeling data, and the second labeling data and the first input data may be included in the third learning data.

In a further embodiment, the step of generating third training data including second labeling data and first input data may be performed when the class included in the first prediction result and the class included in the first labeling data do not correspond to each other. In this case, the first input data may not be included in the third learning data. In this embodiment, if the class included in the first prediction result is “non-combined” and the class included in the first labeling data is “combined” and do not correspond to each other, the first input data corresponding to the first prediction result is 3 May not be used to build training data. Accordingly, the technical effect that the reliability and accuracy of the third learning data can be increased can be derived.

Additional description of the step of generating third learning data including second labeling data and first input data will be described later with reference to FIG. 6 .

In one embodiment, the computing device 100 may train a second prediction model to classify a third number of classes based on the third learning data (S350).

In one embodiment, prediction models may be operated based on any learning method to classify a predetermined number of classes based on training data.

In one embodiment, MHC and peptides capable of binding, or each of MHC and peptides, may be obtained through experimental result data and/or public databases (e.g., IEDB, HLA atlas Ligand, SysteMHC Altas).

Additional description of the step of training the second prediction model (or first prediction model) to classify the third number of classes based on the third learning data will be described later with reference to FIG. 6.

Figure 4 exemplarily shows a method of training a prediction model for binary classifying pMHCs according to the degree of binding according to an embodiment of the present disclosure.

According to one embodiment, a method of training a prediction model that binary classifies pMHCs according to the degree of binding is illustratively shown in FIG. 4. The steps shown in FIG. 4 may be performed by computing device 100.

The first prediction model 420, which is an example of a prediction model according to an embodiment, may be trained using the second learning data 410. The second learning data 410 may be composed of pMHC learning data 412a, 412b identified as combined and pMHC learning data 414a, 414b identified as not combined.

In one embodiment, the pre-training of the first prediction model 420 includes pMHC (Binding pMHC) identified as having occurred through experimental data and pMHC (Non-binding pMHC) identified as having not occurred through experimental data. ) may include supervised learning, semi-supervised learning, unsupervised learning, transfer learning, or reinforcement learning using the correct answer data of ) as the learning data set.

For example, in order to perform the semi-supervised learning of the first prediction model, the computing device 100 labels (412a) Binding pMHC a portion of pMHC identified as having been bound through experimental data (labeled, L). , the part of pMHC identified as having bound through experimental data can be proceeded without labeling it as Binding pMHC (412b) (unlabeled, U). Additionally, in order to perform the semi-supervised learning of the first prediction model, the computing device 100 labels a portion of pMHC identified through experimental data as non-binding pMHC as non-binding pMHC (414a), and performs the semi-supervised learning of the first prediction model. Some of the pMHC identified through data as having not occurred can be processed without labeling them as non-binding pMHC (414b).

In a further embodiment, as part of the training data, labeled training data 412a, 414a, 512a, 514a, 612a, 614a (L in FIGS. 4, 5, and 6) and other parts of the training data, The composition ratio of the unlabeled training data (412b, 414b, 512b, 514b, 612b, 614b) (U in FIGS. 4, 5, and 6) is used to evaluate the performance of the prediction models (420, 520, 640). It may be determined differently. For example, in FIG. 4, the ratio of labeled data (412a, 414a) and unlabeled data (412b, 414b) constituting the second learning data 410 is expressed as 1:2, but the ratio of L and U is not limited to this.

In an additional embodiment, the composition ratio of L and U in the learning data is set to a different composition ratio of L and U by evaluating the performance of the prediction model, so that the accuracy of the prediction model can be further increased.

In one embodiment, the classification result 430 output by the first prediction model 420 may include a plurality of classes related to the binding of a peptide and MHC. For example, a plurality of classes may be related to the degree of binding between the peptide constituting pMHC and MHC.

In a further embodiment, the classification result 430 output by the first prediction model 420 includes a first class 432 and a second class 434 included in a second number (e.g., two) of classes. It can be done with

In a further embodiment, the first class 432 and the second class 434 may be classified based on a first score corresponding to the first class 432 and a second score corresponding to the second class 434. there is. In one example, the first score identifies binding as a result of MS analysis (e.g., IC50 ≤ 500nM) and the second score identifies binding as a result of MS analysis (e.g., IC50 > 500nM) to classify the classes ( 432, 434) can be distinguished.

FIG. 5 exemplarily illustrates a method of generating new classes by updating labeling data based on pre-existing labeling data and a prediction result output through a prediction model according to an embodiment of the present disclosure.

In one embodiment, new classes are created by updating labeling data based on the first prediction result 532 classified through the first prediction model 520 and the first labeling data 512a, 512b, 514a, and 514b. You can.

In one embodiment, based on a combination of the first number of classes 532 and the second number of classes 430, first labeling data 53a, 53b corresponding to the first learning data 510 may be updated with second labeling data 54a, 54b, 54c, and 54d corresponding to the third number of classes 540.

In one embodiment, the number in the third number of classes 540 may mean a natural number of 3 or more. In a further embodiment, the second labeling data 54a, 54b, 54c, 54d corresponding to a third number (e.g. four) of classes 540 refers to quantified binding of pMHC, for example. Class 1', which means quantified weak binding of pMHC, Class 2', which means quantified weak non-binding of pMHC, Class 3', which means quantified weak non-binding of pMHC, and Class 4' which means quantified non-binding of pMHC. can be responded to. In this example, the first number of classes 532 may include two classes representing association or non-association obtained according to a low-confidence test, and the second number of classes 430 may include a high-confidence test. It may contain two classes representing the binding or non-binding obtained according to the test. The output of the first prediction model pretrained with the second number of classes 430 will include the second number of classes 430, and the first learning input with this second number of classes 430 By comparing the first number of classes 532 labeled in the data, new types of classes 540 to be labeled in the first input data may be created. In this example, according to the combination of two classes (53a, 53b) corresponding to the first number of classes (532) and two classes (432, 434) corresponding to the second number of classes (430) A total of four classes can be created with the third number of classes 540.

In one embodiment, the first number of classes 532 included in the output of the first prediction model 520 are output as combined pMHC 53a and non-bound pMHC 53b according to binary classification.

In a further embodiment, the creation of a new class 500 may be performed, for example, when the pMHC training data 512a, 512b identified as combined in the first training data 510 are predicted to be a combined pMHC 53a. , may include the process of creating a new class 1' (54a). In addition, when the pMHC learning data (512a, 512b) identified as combined in the first learning data (510) is predicted to be an unbound pMHC (53b), a process of generating a new class 2' (54a) may be included. You can.

As a further example, the creation of a new class 500 may generate new class 3' when the pMHC training data 514a, 514b identified as not combined in the first training data 510 are predicted to be a combined pMHC 53a. It may include a process of generating (54c). In addition, when the pMHC learning data (514a, 514b) identified as non-binding in the first learning data (510) is predicted to be non-binding pMHC (53b), a process of generating a new class 4' (54d) is included. can do. In this way, the creation of new classes 500 can be class 1', class 2', class 3', or class 4', representing binding pMHC, weakly binding pMHC, weakly unbinding pMHC, or nonbinding pMHC, respectively (54a, 54b). , 54c, and 54d) can be created.

According to one embodiment, the first learning data 610 to which labeling based on the four new classes 540 and 620 is applied may constitute third learning data 630, and this third learning data Based on 630, the first prediction models 420 and 520 may be retrained or the second prediction model 640 may be trained. In this way, the accuracy of labeling results according to the results of tests with relatively low accuracy or reliability can be increased, and the technical effect that a sufficient amount of learning data can be acquired and generated can be achieved.

6 exemplarily illustrates a method of training a prediction model based on updated labeling data and training data according to an embodiment of the present disclosure.

In one embodiment, based on third training data 630 including new classes 620 and re-labeling 63 data based on first training data 610 and first input data. A second prediction model 640 may be trained.

In one embodiment, the input data, labeling data, and/or relabeling data pool included in the third learning data 630 is limited to the first learning data 610, as shown in FIG. 6. It doesn't work. For example, the first learning data 610 for configuring the data included in the third learning data 630 is the first learning data 510 used to train the first prediction model 520 of FIG. 5. It may include different types of public databases and/or experimental result data. The second prediction model 640 can be trained by configuring the third learning data 630 in this way (632a, 632b, 634a, 634b).

In one embodiment, the second prediction model 640 may output a prediction result including a third number of classes 650. For example, the third number of classes 650 may have four classes. Each class is classified into a lower-numbered class (e.g., class 1 (65a)) the higher the probability that the peptide and MHC are bound, and the lower the probability that the peptide and MHC are unbound, the higher-numbered class (e.g., class N (65c), where N is a natural number). In a further example, each class is classified into a lower numbered class (e.g., class 1 (65a)) as the binding between the peptide and the MHC is stronger, and into a higher numbered class (e.g., class N) as the binding between the peptide and the MHC is weaker. (65c), where N is a natural number). These classes 65a, 65b, and 65c can be numbered in a way that is advantageous for driving a prediction model of the computer device 100, and are not limited to numbers, vectors, matrices, etc.

In one embodiment, in the inference process of the prediction model 640, the new classes 620 of the pMHC classification result for the first training data and the pMHC classification result 532 for the second training data are combined or other By inputting it into the model, the pMHC classification result 650 corresponding to the learning data set can be output. This prediction model training method can produce multifaceted and effective pMHC classification results.

7 illustratively illustrates a tokenization process according to one embodiment of the present disclosure.

In one embodiment, preprocessing of first input data 710 may include tokenization process 720 and segmentation process 730. The tokenization process 720 may include tokenizing amino acid sequences in single or multiple units. The segmenting process 730 is a process of combining the tokenized token groups 712 and 714 and/or separating the amino acid sequence tokens of the peptide and the amino acid sequence tokens of the MHC among the amino acid sequences constituting the pMHC. It can mean a process. In one embodiment, the training data set 740 is data generated based on the tokenization process 720 and the segmentation process 730 and can be used for training a prediction model.

In one embodiment, tokenization process 720 may include a process of generating tokens having a length corresponding to each of the different length amino acid sequences by analyzing the frequency of occurrence for each of the different length amino acid sequences. there is. Here, the length of the amino acid sequence may correspond to the number of amino acids included in the amino acid sequence. For example, if the length of the amino acid sequence is 4, the number of amino acids included in the amino acid sequence may be 4. Input data subject to labeling can be labeled and learned in units of tokens according to the tokenization result.

In one example, the tokenization process may include a token grouping process. That is, tokenization of the first input data may be performed based on tokens generated by the token grouping process.

In one embodiment, the frequency of appearance may represent a value that quantitatively represents the probability of a specific amino acid sequence being found within the amino acid sequence of one peptide. In one embodiment, the frequency of appearance may quantitatively represent the ratio of the number of times a specific amino acid sequence is found in all peptides to the total number of peptides.

In one embodiment, the tokenization process includes generating a first group of tokens for a first set of amino acid sequences obtained by analyzing the frequency of occurrence for amino acid sequences having a first length, and the first token Comprising a second set of amino acid sequences obtained by analyzing the frequency of occurrence of amino acid sequences having a second length shorter than the first length, with the first set of amino acid sequences included in the group excluded. It may include the step of generating a second token group. Here, tokens included in the first token group may have a first length and tokens included in the second token group may have a second length.

In one embodiment, the segmenting process may refer to a process of combining grouped groups. In one example, the segmentation process may include combining tokens to create a training data set. The segmenting process may include, for example, segment embedding.

In one embodiment, the segmentation process comprises tokens corresponding to peptides of the first input data and tokens corresponding to the MHC of the first input data with a separator token interposed therebetween. 1 Learning data can be generated.

In one embodiment, after training of the prediction model 750 is completed, the prediction model 750 may receive first input data 710 corresponding to the peptide and MHC and output classification results according to the degree of pMHC binding. there is. In an additional embodiment, after the training of the prediction model 750 is completed, the prediction model 750 receives first input data 710 corresponding to the peptide and MHC, and the training data of another prediction model - the training data May include classes related to binding of labeling data and pMHC - Input data to be included in can be output. In an additional embodiment, after training of the prediction model 750 is completed, the prediction model 750 receives first input data 710 corresponding to peptides and MHC to change the classification criteria of the prediction model 750. Alternatively, in order to change the learning method, input data to be included in the learning data for retraining the prediction model 750 can be output.

In one embodiment, the first input data 710 may include an amino acid sequence corresponding to a peptide and/or an amino acid sequence corresponding to MHC.

In one embodiment, tokenization process 720 may generate tokens 712 and 714 corresponding to input data 710. In one embodiment, tokenization process 720 generates a first token 712 corresponding to a peptide in first input data 710 and a second token corresponding to an MHC in first input data 710 ( 714) can be created.

In one embodiment, tokenization process 720 includes a process for splitting input data into smaller units (e.g., shorter sequences of amino acid delimiters), where the split units may correspond to tokens.

Segmenting process 730 according to an embodiment of the present disclosure may include a process of creating a training data set 740 based on tokens.

In one embodiment, segmenting process 730 may generate a training data set 740 that includes first input data 710 corresponding to peptides and MHC. In this learning data set 740, some or all of the classification results corresponding to the first input data 710 may be reused as the correct answer data set.

In a further embodiment, a training data set 740 is generated through the segmentation process 730 in an integrated form, for example, by concatenating amino acid sequences corresponding to peptides and amino acid sequences corresponding to MHC. It can be.

In one embodiment, the amino acid sequences included in the first input data 710 are various numbers (K), such as 7 amino acids, 8 amino acids, 9 amino acids, 10 amino acids, or 11 amino acids. ) may be composed of amino acids. These specific lengths of amino acids may be referred to as an amino acid set. For example, assuming that the total number of amino acids is 20, amino acid sequences with a length of 4 may exist in a total number of 20 to the power of 4.

During the tokenization process 720 for each set of amino acids whose amino acid sequence has a length of K according to an embodiment of the present disclosure, the frequency of appearance in the peptide or MHC included in the first input data 710 is calculated. You can decide. The process of determining how many times a set of amino acids of a certain length appears or exists within the range of amino acid sequences included in such peptides or MHC may be referred to as frequency of occurrence analysis. In an additional embodiment, specific amino acid sequences among amino acid sequences included in the first input data 710 may be tokenized through frequency of occurrence analysis.

In one embodiment, tokenization process 720 may obtain the peptide list from an external database or a database included in computing device 100. In a further embodiment, computing device 100 may determine, for each set of amino acids whose amino acid sequence length is K, a first frequency of occurrence within the peptide list. In a further embodiment, the computing device 100 may determine, among sets of amino acids whose amino acid sequences have a length of K, a first set of amino acids whose first frequency of occurrence is equal to or greater than a first predetermined threshold. In a further embodiment, computing device 100 may determine, for each set of amino acids whose amino acid sequence length is K-1, a second frequency of occurrence within the peptide list. In a further embodiment, computing device 100 may determine, among sets of amino acids whose amino acid sequences have a length of K-1, a second set of amino acids whose second frequency of occurrence is equal to or greater than a second predetermined threshold. In a further embodiment, computing device 100 may perform tokenization with a length of K for the first set of amino acids and perform tokenization with a length of K-1 for the second set of amino acids. Here, K is a natural number of 2 or more, and the first threshold value and the second threshold value may have the same or different values.

In one embodiment, the analysis of the frequency of occurrence may not be performed on the MHC sequence included in the first input data 710, but only on the peptide sequence included in the first input data 710. This may be because the MHC amino acid sequence between MHC types (eg, HLA-A01, HLA-B03, HLA-A02) may have little specific variation or may not be continuous. According to this embodiment, improved prediction results can be generated by performing the tokenization process 720 without using the frequency of occurrence in the MHC amino acid sequence as a factor.

In one embodiment, tokenization process 720 may obtain the peptide list from an external database or a database included in computing device 100. The tokenization process 720 includes in the token list 712, 714 the M+1th set of amino acids whose frequency of occurrence in the peptide list is greater than or equal to a predetermined threshold among the sets of amino acids whose amino acid sequences are N-M in length. And the token lists 712 and 714 can be constructed by removing the M+1 amino acid set from the peptide list. After constructing the token lists 712 and 714, the tokenization process 720 may increment the value of M by 1 and determine whether a termination condition has been satisfied. The tokenization process 720 re-performs the steps of constructing the token lists 712 and 714 if the termination condition is not satisfied, and if the termination condition is satisfied, adds to each of the amino acid sets included in the token lists 712 and 714. Tokenization can be performed by generating a token corresponding to each set of amino acids with the corresponding amino acid sequence length. Here, M is an integer greater than 0, and N-M can be a natural number greater than 2.

In a further example, the termination condition may include conditions for determining whether to further perform the frequency analysis while reducing the length of the amino acid sequence or to end the frequency analysis here.

For example, the termination condition may include a first termination condition corresponding to N-M≤1. If the above termination condition is satisfied based on the current values of N and M, the computing device 100 may terminate the frequency of occurrence analysis without increasing the value of M. As another example, the termination condition may include a second termination condition corresponding to N-M≤2.

In a further embodiment, the threshold used in the frequency of occurrence analysis may be varied to have a negative correlation with the magnitude of the value of N-M. For example, when the value of N-M is large, the size of the threshold may be relatively low.

The prediction model in this disclosure may mean an artificial intelligence-based generation model. In one example, the prediction model may be a Recurrent Neural Network (RNN), Long Short Term Memory (LSTM) network, Bidirectional Long Short Term Memory (BiLSTM) network, Bidirectional Encoder Representations from Transformers (BERT), Gated Recurrent Unit (GRU), or BiGRU. (Bidirectional Gated Recurrent Unit) may be included.

8 is a schematic diagram of a computing environment according to one embodiment of the present disclosure.

A component, module, or unit in the present disclosure includes routines, procedures, programs, components, data structures, etc. that perform a specific task or implement a specific abstract data type. Additionally, one of ordinary skill in the art will understand that the methods presented in this disclosure can be used in uni-processor or multiprocessor computing devices, minicomputers, mainframe computers, as well as personal computers, handheld computing devices, microprocessor-based or programmable consumer electronics, etc. ( It will be fully appreciated that each of these may be implemented with other computer system configurations, including those capable of operating in conjunction with one or more associated devices.

Embodiments described in this disclosure can also be practiced in distributed computing environments where certain tasks are performed by remote processing devices that are linked through a communications network. In a distributed computing environment, program modules may be located in both local and remote memory storage devices.

Computing devices typically include a variety of computer-readable media. Computer-readable media can be any medium that can be accessed by a computer, and such computer-readable media includes volatile and non-volatile media, transitory and non-transitory media, removable and non-transitory media. Includes removable media. By way of example, and not limitation, computer-readable media may include computer-readable storage media and computer-readable transmission media.

Computer-readable storage media refers to volatile and non-volatile media, transient and non-transitory media, removable and non-removable, implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data. Includes media. Computer readable storage media may include RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital video disk (DVD) or other optical disk storage, magnetic cassette, magnetic tape, magnetic disk storage or other magnetic storage. This includes, but is not limited to, a device, or any other medium that can be accessed by a computer and used to store desired information.

A computer-readable transmission medium typically implements computer-readable instructions, data structures, program modules, or other data on a modulated data signal, such as a carrier wave or other transport mechanism. Includes all information delivery media. The term modulated data signal refers to a signal in which one or more of the characteristics of the signal have been set or changed to encode information within the signal. By way of example, and not limitation, computer-readable transmission media includes wired media such as a wired network or direct-wired connection, and wireless media such as acoustic, RF, infrared, and other wireless media. Combinations of any of the above are also intended to be included within the scope of computer-readable transmission media.

An example environment 2000 is shown that implements various aspects of the invention, including a computer 2002, which includes a processing unit 2004, a system memory 2006, and a system bus 2008. do. Computer 200 herein may be used interchangeably with computing device. System bus 2008 couples system components, including but not limited to system memory 2006, to processing unit 2004. Processing unit 2004 may be any of a variety of commercially available processors. Dual processors and other multiprocessor architectures may also be used as processing units 2004.

System bus 2008 may be any of several types of bus structures that may further be interconnected to a memory bus, peripheral bus, and local bus using any of a variety of commercial bus architectures. System memory 2006 includes read only memory (ROM) 2010 and random access memory (RAM) 2012. The basic input/output system (BIOS) is stored in non-volatile memory (2010), such as ROM, EPROM, and EEPROM, and is a basic input/output system (BIOS) that helps transfer information between components within the computer (2002), such as during startup. Contains routines. RAM 2012 may also include high-speed RAM, such as static RAM for caching data.

Computer 2002 may also read from or use an internal hard disk drive (HDD) 2014 (e.g., EIDE, SATA), magnetic floppy disk drive (FDD) 2016 (e.g., removable diskette 2018). (for writing to), SSDs, and optical disk drives (2020) (e.g., for reading CD-ROM disks (2022) or for reading from or writing to other high-capacity optical media, such as DVDs). Includes. The hard disk drive 2014, magnetic disk drive 2016, and optical disk drive 2020 are connected to a system bus 2008 by a hard disk drive interface 2024, magnetic disk drive interface 2026, and optical drive interface 2028, respectively. ) can be connected to. The interface 2024 for implementing an external drive includes, for example, at least one or both of Universal Serial Bus (USB) and IEEE 1394 interface technologies.

These drives and their associated computer-readable media provide non-volatile storage of data, data structures, computer-executable instructions, and the like. For the computer 2002, drive and media correspond to storing any data in a suitable digital format. Although the description of computer-readable storage media above refers to removable optical media such as HDDs, removable magnetic disks, and CDs or DVDs, those skilled in the art will also recognize removable optical media such as zip drives, magnetic cassettes, flash memory cards, cartridges, etc. It will be appreciated that other types of computer-readable storage media may also be used in the exemplary operating environment and that any such media may contain computer-executable instructions for performing the methods of the invention. .

A number of program modules may be stored in the drive and RAM 2012, including an operating system 2030, one or more application programs 2032, other program modules 2034, and program data 2036. All or portions of the operating system, applications, modules and/or data may also be cached in RAM 2012. It will be appreciated that the invention may be implemented on various commercially available operating systems or combinations of operating systems.

A user may input commands and information into the computer 2002 through one or more wired/wireless input devices, such as a pointing device such as a keyboard 2038 and a mouse 2040. Other input devices (not shown) may include microphones, IR remote controls, joysticks, game pads, stylus pens, touch screens, etc. These and other input devices are connected to the processing unit 2004 through an input device interface 2042, which is often connected to the system bus 2008, but may also include a parallel port, an IEEE 1394 serial port, a game port, a USB port, an IR interface, It can be connected by other interfaces, etc.

A monitor 2044 or other type of display device is also connected to system bus 2008 through an interface, such as a video adapter 2046. In addition to the monitor 2044, computers typically include other peripheral output devices (not shown) such as speakers, printers, etc.

Computer 2002 may operate in a networked environment using logical connections to one or more remote computers, such as remote computer(s) 2048, via wired and/or wireless communications. Remote computer(s) 2048 may be a workstation, server computer, router, personal computer, portable computer, microprocessor-based entertainment device, peer device, or other conventional network node, and generally refers to computer 2002. For simplicity, only memory storage device 2050 is shown, although it includes many or all of the components described. The logical connections depicted include wired/wireless connections to a local area network (LAN) 2052 and/or a larger network, such as a wide area network (WAN) 2054. These LAN and WAN networking environments are common in offices and companies and facilitate enterprise-wide computer networks, such as intranets, all of which can be connected to a worldwide computer network, such as the Internet.

When used in a LAN networking environment, computer 2002 is connected to local network 2052 through wired and/or wireless communications network interfaces or adapters 2056. Adapter 2056 may facilitate wired or wireless communication to LAN 2052, which also includes a wireless access point installed thereon for communicating with wireless adapter 2056. When used in a WAN networking environment, the computer 2002 may include a modem 2058, connected to a communication server on the WAN 2054, or other means of establishing communication over the WAN 2054, such as via the Internet. has Modem 2058, which may be internal or external and a wired or wireless device, is coupled to system bus 2008 via serial port interface 2042. In a networked environment, program modules described for computer 2002, or portions thereof, may be stored in remote memory/storage device 2050. It will be appreciated that the network connections shown are exemplary and that other means of establishing a communications link between computers may be used.

Computer 1602 may be associated with any wireless device or object deployed and operating in wireless communications, such as a printer, scanner, desktop and/or portable computer, portable data assistant (PDA), communications satellite, wirelessly detectable tag. Performs actions to communicate with any device or location and telephone. This includes at least Wi-Fi and Bluetooth wireless technologies. Accordingly, communication may be a predefined structure as in a conventional network or may simply be ad hoc communication between at least two devices.

It is to be understood that the specific order or hierarchy of steps in the processes presented is an example of illustrative approaches. It is to be understood that the specific order or hierarchy of steps in processes may be rearranged within the scope of the present disclosure, based on design priorities. The method claims of this disclosure provide elements of the various steps in a sample order but are not meant to be limited to the specific order or hierarchy presented.

Claims (1)

As a method of training an artificial intelligence-based prediction model performed by a computing device,
Obtaining first learning data including first input data corresponding to a peptide and a Major Histocompatibility Complex (MHC) and first labeling data corresponding to the first input data - the first learning data 1 Labeling data corresponds to a first number of classes related to binding of peptides and MHC -;
Using a first prediction model pre-trained based on the second training data to classify a second number of classes related to the binding of a peptide and an MHC, from the first training data, a first prediction model for the second number of classes is generated. 1 Obtaining a prediction result;
Based on the first prediction result and the first labeling data, updating the first labeling data of the first learning data with second labeling data corresponding to a third number of classes related to the combination of peptide and MHC. step;
generating third learning data including the second labeling data and the first input data; and
Based on the third training data, training a second prediction model to classify the third number of classes;
Including,
method.