Method and computer program for predicting neoantigens using peptide and HLA allele sequences

Abstract

The present disclosure discloses a method for predicting neoantigens using peptide sequences and HLA allele sequences, the method including: receiving as input peptide sequences and HLA allele sequences extracted from a target cancer tissue; acquiring T cell activity data from the peptide sequences, inputting the T cell activity data into an immune response prediction model, and outputting a first predicted value that predicts the immune response of the peptide sequence; acquiring binding data from the HLA allele sequences, inputting the binding data into a binding response prediction model, and outputting a second predicted value that predicts the binding response of the peptide sequence and the HLA allele sequence; inputting the T cell activity data and the binding data into an immune resistance prediction model, and outputting a third predicted value that predicts the immune resistance of the target cancer tissue; and generating neoantigen information for the target cell using the T cell activity data and the first to third predicted values.

Description

[Technical field]

[0001] The present disclosure relates to methods and computer programs for predicting neoantigens using peptide sequences and HLA allele sequences. [Background technology]

[0002] With the development and advancement of new anti-cancer drugs, 3rd generation immune anti-cancer drugs have recently been in the limelight after 1st generation anti-cancer drugs such as chemo-anti-cancer drugs and 2nd generation targeted anti-cancer drugs. In particular, 3rd generation immune anti-cancer drugs have the advantage of having significantly fewer side effects than previous generation anti-cancer drugs because they are a treatment strategy that utilizes the patient's own immune system. However, despite these advantages, there is a limitation that only patients who express marker genes such as PD-L1 and show microsatellite instability (MSI-H) can establish a treatment strategy using immune anti-cancer drugs. Due to these limitations, it is necessary to establish a strategy to treat patients who are difficult to administer existing anti-cancer drugs, and one of the alternatives proposed is a cancer vaccine using neoantigens. Each patient's cancer tissue has mutations that are not found in normal tissues, and the core strategy of a cancer vaccine is to use peptides derived from these mutations as neoantigens so that the patient's immune system can recognize and attack the neoantigens. In this process, the first step is to ensure stable binding between the mutation-derived peptide and the patient-specific HLA allele, and the second step is to confirm whether the mutation-derived peptide has immunogenicity that stimulates the patient's immune system. In particular, in order to fully reflect immunogenicity, all steps in which immunogenicity occurs must be replicated and the main features must be extracted, but in this process, there may be steps that are missed or lost, which may act as a limiting point in future immunogenicity predictions.

[0003] Therefore, this technology attempts to realize a strategy to prevent leakage in the immune process and extract key features based on combination data of peptides with immunogenicity and MHC (HLA in the case of humans) sequences known up to now. In addition, it attempts to derive neoantigens applicable to patients by modeling the binding strength with HLA as a neoantigen, the immunogenicity of the peptide sequence itself, and immune resistance that has binding strength but does not induce immunity. The above-mentioned background art is technical information that the inventor possessed for the derivation of the present invention or acquired in the process of derivation of the present invention, and cannot necessarily be said to be publicly known technology disclosed to the general public prior to the filing of the present invention. Summary of the Invention [Problem to be solved by the invention]

[0004] The present invention is based on the above-mentioned need, and aims to predict that peptide sequences derived from mutations present in a patient's cancer tissue will bind to patient-specific HLA while ultimately exhibiting immunogenicity through a series of immune processes, and based on this, to determine neoantigens that can be used in cancer vaccines suitable for cancer patients. [Means for solving the problem]

[0005] A method for predicting a neoantigen using a peptide sequence and an HLA allele sequence according to an embodiment of the present invention includes the steps of receiving a peptide sequence and an HLA allele sequence extracted from a target cancer tissue as an input, acquiring T cell activity data from the peptide sequence, inputting the T cell activity data into an immune response prediction model, and outputting a first predicted value for predicting the immune response of the peptide sequence, acquiring binding data from the HLA allele sequence, inputting the binding data into an immune response prediction model, and outputting a second predicted value for predicting the affinity of the peptide sequence and the HLA allele sequence, inputting the T cell activity data and the binding data into an immune resistance prediction model, and outputting a third predicted value for predicting the immune resistance of the target cancer tissue, and generating neoantigen information for the target cell using the T cell activity data and the first to third predicted values.

[0006] At least one of the immune prediction model, the binding prediction model, and the immune resistance prediction model can be trained by a machine learning algorithm based on a training dataset including peptide sequences and HLA allele sequences present in a plurality of target cancer tissues.

[0007] The subject cancer tissue can comprise cells engineered to express a single MHC class I or class II allele.

[0008] The subject cancer tissue can include human cells obtained or derived from multiple patients.

[0009] The target cancer tissue may include fresh or frozen tumor cells obtained from multiple patients.

[0010] The target cancer tissue may include fresh or frozen tissue cells obtained from multiple patients.

[0011] The target cancer tissue can contain a peptide identified using a T cell assay.

[0012] The training dataset may include at least one of data related to a protein sequence associated with the subject cancer tissue, data related to an MHC peptide sequence associated with the subject cancer tissue, binding data between a peptide associated with the subject cancer tissue and an HLA allele, data related to a transcriptome associated with the subject cancer tissue, and data related to a genome associated with the subject cancer tissue.

[0013] The immune prediction model may be a model trained using T cell activity data from a peptide sequence as input and the immune activity of the peptide sequence as output.

[0014] The binding prediction model can be a model trained using binding data from an HLA allele sequence and a peptide sequence as input, and using the binding properties of the peptide sequence and the HLA allele sequence as output.

[0015] The immune tolerance prediction model may be a model trained using T cell activity data from peptide sequences and HLA allele sequences and binding data from HLA allele sequences and peptide sequences as inputs, and immune tolerance between peptide sequences and HLA allele sequences as output.

[0016] A computer program according to the embodiments of the present invention may be stored on a medium for causing a computer to execute any one of the methods according to the embodiments of the present invention.

[0017] Additionally, other methods for implementing the present invention, other systems, and computer readable recording media having computer programs recorded thereon for carrying out the methods are also provided.

[0018] Other aspects, features, and advantages will become apparent from the following drawings, claims, and detailed description of the invention. Effect of the Invention

[0019] According to one embodiment of the present invention as described above, the binding strength between a peptide sequence and an HLA allele sequence contained in a cancer tissue as well as the immunological activity of the peptide sequence can be measured, and neoantigens in the cancer tissue can be determined based on the measured immunological activity. [Brief description of the drawings]

[0020] [Figure 1] FIG. 1 is a block diagram of a neoantigen determining apparatus 100 according to an embodiment of the present invention. [Diagram 2] FIG. 1 is an exemplary diagram showing HLA class I type information contained in the cells of a Korean person. [Diagram 3]FIG. 1 is a block diagram of a neoantigen determining apparatus 100 according to an embodiment of the present invention. [Figure 4] FIG. 2 is a diagram for explaining input data and output data of the neoantigen determination apparatus 100. [Diagram 5] 1 is a diagram showing an example of an embodiment of a neoantigen determination system according to an embodiment of the present invention; [Figure 6] 1 is a diagram showing an example of an embodiment of a neoantigen determination system according to an embodiment of the present invention; [Figure 7] 1 is a diagram showing an example of an embodiment of a neoantigen determination system according to an embodiment of the present invention; [Figure 8] FIG. 1 is a block diagram of a learning server 10 that trains an immune prediction model, a binding prediction model, an immune resistance prediction model, etc. [Figure 9] 1 is an exemplary diagram of an immune prediction model, a binding prediction model, and an immune resistance prediction model. FIG. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0021] The configuration and operation of the present invention will be described in detail below based on the embodiments of the present invention shown in the accompanying drawings.

[0022] The present invention can be modified in various ways and can have various embodiments, but a specific embodiment will be illustrated in the drawings and described in detail. The advantages and features of the present invention, as well as methods for achieving them, will become apparent from the following detailed description of the embodiments with reference to the drawings. However, the present invention is not limited to the embodiments disclosed below, and can be embodied in various forms.

[0023] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. In the description with reference to the drawings, the same or corresponding elements are designated by the same reference numerals, and redundant description thereof will be omitted.

[0024] In the following embodiments, the terms "first", "second", etc. are not used in a limiting sense but are used to distinguish one component from another component.

[0025] In the following examples, singular expressions include plural expressions unless the context clearly dictates otherwise.

[0026] In the following examples, terms such as "comprise" or "have" mean the presence of features or components described in the specification, and do not preclude the possibility that one or more other features or components may be added.

[0027] In the drawings, the size of components may be exaggerated or reduced for the convenience of explanation. For example, the size and thickness of each component shown in the drawings are arbitrarily shown for the convenience of explanation, and the present invention is not necessarily limited to those shown in the drawings.

[0028] In alternative implementations of an embodiment, the order of certain steps may be performed differently than described, for example, two steps that are shown in succession may be performed substantially simultaneously or may be performed in the reverse order of the steps described.

[0029] Here, the target cancer tissue refers to a tissue that is the subject of an experiment. For example, the target cancer tissue is a cancer tissue in which an antigen capable of inducing an immune response is to be detected. Preferably, the target cancer tissue is a tumor cell or a collection of cancer cells.

[0030] Here, mutation refers to any phenomenon in which the base sequence A (adenine), T (thiamine), G (guanine), and C (cytosine) of the gene that carries the genetic information in each living organism is altered to be different from the original genetic information of the species. Such mutations cause small-scale or large-scale structural changes, and small-scale mutations include point mutations that appear when a single base sequence is changed, and there are also mutations in which additional base sequences are inserted or deleted. Mutations that occur on a large scale and affect structure include gene duplication, gene deletion, chromosomal inversion, epilepsy deletion, chromosomal translocation, and loss of heterozygosity.

[0031] Mutations are broadly divided into germ cell mutations and somatic mutations, depending on the type of cell they occur in. Somatic mutations are genetic mutations that occur in somatic cells, and can be caused by genetic mutations or chromosomal abnormalities.

[0032] The occurrence of such mutations can cause changes in the function of the protein produced from the gene, and a specific function can be lost or another function can be activated. Such changes in protein function can cause or accelerate the onset of cancer, so such mutations may be directly or indirectly closely related to the onset and progression of cancer.

[0033] As mentioned above, the base sequence in DNA, which holds the genetic information of an organism, consists of A, T, G, and C. When three of these base sequences are arranged in a row, they form a code for forming a specific amino acid, and when many such codes are assembled, they can be converted into a protein. Amino acids consist of alanine (Ala), cysteine ​​(Cys), aspartic acid (Asp), glutamic acid (Glu), phenylalanine (Phe), glycine (Gly), histidine (His), isoleucine (Ile), lysine (Lys), leucine (Leu), methionine (Met), asparagine (Asn), pyrrolysine (Ply), proline (Pro), glutamine (Gln), arginine (Arg), serine (Ser), threonine (Thr), selenocysteine ​​(Sec), valine (Val), tryptophan (Trp), and tyrosine (Tyr).

[0034] A peptide may refer to a peptide or polypeptide consisting of an amino acid sequence. In living organisms, there is an immune system for removing foreign substances that are not derived from various genetic information, and among the foreign peptides, there are immunogenic peptides that can cause an immune response. Mutations that occur differently from the original genetic information during the development of cancer also produce such immunogenic peptides, and these peptides can bind to MHCI protein through a series of processes in the immune system. Furthermore, the immunogenic peptide may have a mutated amino acid sequence, and the length of the amino acid may be 25 or less, but is not limited thereto, and may be of various lengths.

[0035] Neoantigen refers to a peptide that induces an immune response. That is, neoantigen can be an immunogenic peptide. Neoantigen can be induced by tumor cell specific mutation and can be represented by tumor cell epitopes. In the following, for the sake of simplicity, immunogenic peptides are named as neoantigens.

[0036] Here, T cell activity data refers to data measuring the immune response that occurs when a specific peptide sequence binds to a specific HLA allele and is stimulated, and is obtained as data such as intracellular cytokine expression values ​​and immune cell-specific activity marker expression values ​​detected by immunogenicity measurement experimental methodologies including multimer / tetramer and ELISPOT, and the result values ​​can be classified into "Positive", "Positive-High", "Positive-Low", "Positive-Intermediate", and "Negative".

[0037] The neoantigen determination device according to the embodiment of the present invention can analyze the peptide sequence of the target cancer tissue and the HLA allele sequence of the patient, and determine a specific peptide of the target cancer tissue as a neoantigen to be used in treating the target cancer tissue. It can determine neoantigens suitable as antigens from peptides contained in the target cancer tissue. Antibodies acting on the determined neoantigens can be searched for and used in treating the target cancer tissue of the patient.

[0038] FIG. 1 is a block diagram of a neoantigen determining apparatus 100 according to an embodiment of the present invention.

[0039] The neoantigen determination apparatus 100 is an apparatus for determining neoantigens in order to treat a disease present in a cancer tissue based on the genomic data of the cancer tissue.

[0040] The genome data input unit 110 can receive peptide sequences and HLA allele sequences extracted from cancer tissue. The peptide sequences are for one or more peptides contained in the cancer tissue. The peptide sequences can be expressed in a two-dimensional matrix to include sequences for the peptides. The HLA allele sequences can be expressed in a pseudo sequence or full sequence, and can be expressed in an embedding vector of a specific size by a word embedding technique in which 1 to k amino acid units are set as one word, but are not limited thereto, and can be expressed in various forms.

[0041] The genome data input unit 110 can individually calculate T cell activity data of the peptide or binding data between the peptide and the HLA allele based on the peptide sequence and the HLA allele sequence.

[0042] The genome data input unit 110 can measure T cell activity data for a peptide, and calculate T cell activity data for the peptide in the cancer tissue using a table or database in which the measured data is recorded.

[0043] Here, the HLA allele can be input by dividing the HLA allele sequence in units of 1 kmer, regardless of whether it is the entire sequence or a pseudo sequence, and expressing it as a virtual word set.

[0044] The genome data input unit 110 can measure binding data related to the binding strength for all binding relationships between peptides and HLA alleles, and calculate binding data between peptides and HLA alleles of the target cancer tissue using a table or database in which the measured binding data is recorded.

[0045] The immune prediction unit 121 can input a peptide and HLA as T cell activity data and output a predicted value corresponding to the immune response to the peptide. The immune prediction unit 121 can output a first predicted value corresponding to the immune response to the peptide using a model trained on the T cell activity data and the immune response to the peptide. The peptide may be plural or singular. The T cell activity data may include data for a peptide sequence and / or data for an HLA sequence.

[0046] The binding prediction unit 122 receives as input binding data for all binding relationships between peptides and HLA alleles, and can output a second predicted value corresponding to the binding to the peptide.

[0047] The immune resistance prediction unit 123 receives as input binding data for all binding relationships between peptides and HLA alleles, and can output a third predicted value corresponding to immune resistance to the peptide.

[0048] The neoantigen determining unit 130 can output neoantigen information of the target cancer tissue using a determination model trained based on the first to third predicted values. The neoantigen determining unit 130 can output whether the neoantigen has immunological and binding properties that can be used for treatment based on one of the first to third predicted values ​​determined for each peptide.

[0049] Thus, the neoantigen determining apparatus 100 can output whether a neoantigen can be used for treatment, taking into consideration not only the binding characteristics but also the immunological characteristics of tumor cells or cancer cells contained in the target cancer tissue. In addition, the neoantigen determining apparatus 100 can output whether a neoantigen is available, taking into consideration T cell activity data for peptides of the target cancer tissue.

[0050] The neoantigen determining apparatus 100 may be embodied with at least one of a communication unit, an input unit, and an output unit (not shown), but is not limited thereto. The neoantigen determining apparatus 100 may output data, such as whether a neoantigen is a neoantigen, through the output unit. The neoantigen determining apparatus 100 may receive data input and output through the input unit. The neoantigen determining apparatus 100 may include a communication unit and communicate with an external device. At least one of the genome data input unit 110, the immune predictor 121, the binding predictor 122, the immune resistance predictor 123, and the neoantigen determiner 130 of the neoantigen determining apparatus 100 may be embodied as software or hardware. At least one of the immune predictor 121, the binding predictor 122, the immune resistance predictor 123, and the neoantigen determiner 130 may be embodied as a single component.

[0051] The subject cancer tissue is a cell engineered to express a single MHC class I or class II allele. The subject cancer tissue is a human cell obtained or derived from a plurality of patients. The subject cancer tissue includes fresh or frozen tumor cells obtained from a plurality of patients. The subject cancer tissue includes fresh or frozen tumor cells obtained from a plurality of patients. The subject cancer tissue includes a peptide (etc.) identified by T cell assay.

[0052] The neoantigen determining apparatus 100 can train algorithms of the immune predicting unit 121, the binding predicting unit 122, the immune resistance predicting unit 123, and the neoantigen determining unit 130 based on a plurality of target cancer tissues. The neoantigen determining apparatus 100 can train at least one algorithm of the immune predicting unit 121, the binding predicting unit 122, the immune resistance predicting unit 123, and the neoantigen determining unit 130 using data related to the protein sequence of the target cancer tissue, data related to the MHC peptide sequence, binding data between peptides and HLA alleles, data related to the transcriptome related to the target cancer tissue, data related to the genome related to the target cancer tissue, etc.

[0053] At least one of the immune prediction unit 121, the binding prediction unit 122, the immune resistance prediction unit 123, and the neoantigen determination unit 130 is not constructed independently according to the length of the peptide, but can construct an algorithm (model) by recognizing the peptide as a single word regardless of the length. At least one of the immune prediction unit 121, the binding prediction unit 122, the immune resistance prediction unit 123, and the neoantigen determination unit 130 can embody the peptide as a single word using a word embedding technique.

[0054] Training data for at least one algorithm of the immune activity prediction unit 121, the binding activity prediction unit 122, the immune resistance prediction unit 123, and the neoantigen determination unit 130 may also be input regardless of the length of the peptide. The neoantigen determination unit 130 may use an algorithm trained using deep learning.

[0055] The neoantigen determining apparatus 100 can construct a deep learning model that classifies each of the positive (Y) / negative (N) based on the data. The neoantigen determining apparatus 100 can use an additional neural network by fixing weights for the immune predictor 121, the binding predictor 122, the immune resistance predictor 123, and the neoantigen determiner 130. Thus, at least one of the immune predictor 121, the binding predictor 122, the immune resistance predictor 123, and the neoantigen determiner 130 can be embodied using immune data between HLA alleles and peptides in the T cell activity data.

[0056] The epidemic prediction unit 121 may apply a word embedding technique to each amino acid of the peptide. The epidemic prediction unit 121 may extract feature values ​​by applying CNN to the peptide vector acquired by applying the word embedding technique. Here, the feature values ​​may be acquired by learning in various layers such as CNN. The epidemic prediction unit 121 may generate an algorithm by applying a GRU (Gated Recurrent Unit) to the extracted feature values ​​for the peptide vector to train the positive or negative epidemic of each peptide.

[0057] The binding prediction unit 122 may generate vectors by applying a word embedding technique to both the HLA alleles and the peptides, and extract feature values ​​by applying CNN to the HLA allele vectors and the peptide vectors. The binding prediction unit 122 may generate an HLA allele encoder and a peptide encoder by applying the feature values ​​to two neural networks, and generate an algorithm by a process of training positive or negative binding using the HLA allele encoder and the peptide encoder.

[0058] The training data used to generate the algorithm of the immune resistance prediction unit 123 can be divided into positive and negative for immunity. The immune resistance prediction unit 123 can define a peptide that has binding ability to be bound to an HLA allele but does not have immunity, which is a therapeutic ability, as a tolerant peptide, and can define a peptide that has both binding ability and immunity as a non-tolerant peptide.

[0059] FIG. 2 is an example of HLA class I type information contained in the cells of a Korean person.

[0060] FIG. 3 is a block diagram of a neoantigen determining apparatus 100 according to an embodiment of the present invention.

[0061] As shown in FIG. 3, a neoantigen determining apparatus 100 can generate T cell activity data and binding data.

[0062] The immunological prediction unit of the neoantigen determination apparatus 100 receives T cell activity data as an input and can output a first predicted value for immunological activity.

[0063] The binding prediction unit receives the T cell activity data and the binding data as input, and can output a second predicted value for binding.

[0064] The immune resistance prediction unit receives the T cell activity data and the binding data as input, and can output a third predicted value for immune resistance.

[0065] The neoantigen determining device 100 120 may further include a prediction unit for predicting various factors other than the immune activity prediction unit, the binding activity prediction unit, and the immune resistance prediction unit.

[0066] The neoantigen determination device 100 has the first to third predicted values ​​and the T cell activity data as input, and can output either Y or N as to whether the neoantigen can be used in treatment.

[0067] FIG. 4 is a diagram for explaining input data and output data of the neoantigen determination apparatus 100. As shown in FIG.

[0068] According to an embodiment of the present invention, the HLA allele sequence (X1) and peptide sequence (X2) extracted from the target cancer tissue are used as input data, and N / Y corresponding to whether it is a neoantigen is returned as output data.

[0069] Here, the neoantigen determination device 100 can output N / Y corresponding to whether it is a neoantigen using the immune activity prediction model M1, the binding activity prediction model M2, and the immune resistance prediction model M3. Here, the device can output (return) N / Y corresponding to whether it is a neoantigen by further using a model (NN) trained using the first predicted value output through the immune activity prediction model M1, the second predicted value output through the binding activity prediction model M2, and the third predicted value output through the immune resistance prediction model M3 as inputs and outputting N / Y corresponding to whether it is a neoantigen.

[0070] [Table 1]

[0071] As shown in the table, if there is immunity but no binding between the peptide sequence of the target cancer tissue and the HLA allele sequence, or if there is binding but no immunity, it may be output as N whether it is a neoantigen. If there is both immunity and affinity between the peptide sequence of the target cancer tissue and the HLA allele sequence, it may be output as Y whether it is a neoantigen. Figures 5 to 7 are diagrams showing examples of implementations of a neoantigen determination system according to an embodiment of the present invention.

[0072] 5, the neoantigen determining apparatus 100 can receive genomic data on cancer tissue from an external electronic device 200. The neoantigen determining apparatus 100 can transmit output information on whether the neoantigen is in the cancer tissue to the electronic device 200.

[0073] The electronic device 200 is a computer device including one or more processors that stores genomic data for cancer tissue. The electronic device 200 may be a device that outputs genomic data for cancer tissue. The electronic device 200 may be electrically connected to the neoantigen determination device 100 or connected via a network to transmit and receive data.

[0074] The electronic device 200 can obtain and store genomic data for a plurality of cancer tissue samples over a number of times. The neoantigen determining device 100 can sequentially output whether the genomic data received from the electronic device 200 is a neoantigen.

[0075] As shown in FIG. 6, the neoantigen determining apparatus 100 can receive data from and transmit output data to a plurality of electronic devices 201, 202, ..., 20n.

[0076] The neoantigen determining apparatus 100 can receive genomic data from a number of electronic devices 201, 202, ..., 20n, which can be managed by one or more entities.

[0077] As shown in FIG. 7, the neoantigen determining apparatus 100 may output output data through an output unit of one or more terminal devices 301, 302, ..., 30n. The output data may be output through an output unit of the neoantigen determining apparatus 100. The output data may be output through an output unit of one or more terminal devices 301, 302, ..., 30n. The neoantigen determining apparatus 100 may transmit data related to neoantigens to one or more terminal devices 301, 302, ..., 30n to request payment for a certain fee. One or more terminal devices 301, 302, ..., 30n may request neoantigen-related information regarding peptides and HLA alleles contained in cancer tissues. The output data may be output in response to the request.

[0078] FIG. 8 is a block diagram of a learning server 10 that trains an immune prediction model, a binding prediction model, an immune resistance prediction model, and the like.

[0079] The learning server 10 may include a data input unit 11, a first learning unit 12, a second learning unit 13, a third learning unit 14, and a fourth learning unit 15.

[0080] The first learning unit 12 learns and generates an immune prediction model, and learns T cell activity data of peptide sequences or HLA allele sequences and the immune properties of peptide sequences using a training data set. The immune prediction model learned by the first learning unit 12 processes peptide sequences using a word embedding technique, as shown in 12' of FIG. 10, and inputs the processed peptide sequences into layers of CNN, GRU, and NN for learning.

[0081] The second learning unit 13 learns and generates a binding prediction model, and inputs binding data of a peptide sequence or an HLA allele sequence to learn the binding between the peptide sequence and the HLA allele sequence using a training data set. The binding prediction model learned by the second learning unit 13 processes the peptide sequence and the HLA allele sequence using a word embedding technique, respectively, and inputs the processed peptide sequence to the CNN and GRU layers for learning, and inputs the HLA allele sequence to the CNN, CNN, and GRU layers for learning, as shown in 13' of Fig. 9. The binding prediction model further learns another model NN1 using the predicted value of the binding to the peptide sequence and the predicted value of the binding to the HLA allele sequence, and can finally learn to output a predicted value of the immune resistance to the target cancer tissue.

[0082] The third learning unit 14 learns and generates an immune resistance prediction model, and inputs a peptide sequence and an HLA allele sequence to learn immune resistance between the peptide sequence and the HLA allele sequence using a training data set. The immune resistance prediction model learned by the third learning unit 14 processes the peptide sequence and the HLA allele sequence using a word embedding technique, respectively, and inputs the processed peptide sequence to the CNN and GRU layers for learning, and inputs the HLA allele sequence to the CNN, CNN, and GRU layers for learning, as shown in 14' of Fig. 10. The immune resistance prediction model further learns another model NN2 using the predicted value of immune resistance to the peptide sequence and the predicted value of immune resistance to the HLA allele sequence, and can finally learn to output a predicted value of immune resistance to the target cancer tissue.

[0083] The learning server 10 can transmit the learning models generated by the first to third learning units 12, 13, and 14 to the neoantigen determination apparatus 100. This allows the algorithms of the immune predictor 121, the binding predictor 122, and the immune resistance predictor 123 of the neoantigen determination apparatus 100 to be periodically updated. The above-described apparatus is realized by hardware components, software components, and / or a combination of hardware components and software components. For example, the apparatus and components described in the embodiments can be embodied from one or more general-purpose or special-purpose computers, such as a processor, a controller, an arithmetic logic unit (ALU), a digital signal processor, a microcomputer, a field programmable gate array (FPGA), a programmable logic unit (PLU), a microprocessor, or any other device capable of executing and responding to instructions. The processing device can execute an operating system (OS) and one or more software applications that run on the operating system. A processing device may also access, store, manipulate, process, and generate data in response to the execution of software. For ease of understanding, a single processing device may be described, but those skilled in the art will appreciate that a processing device may include multiple processing elements and / or multiple types of processing elements. For example, a processing device may include multiple processors or a processor and a controller. Other processing configurations, such as parallel processors, are also possible.

[0084] The software may include a computer program, code, instructions, or a combination of one or more of these, and may configure or instruct a processing device to operate as desired, either independently or collectively. The software and / or data may be permanently or temporarily embodied in any type of machine, component, physical device, virtual device, computer storage medium or device, or transmitted signal wave, to be interpreted by the processing device or to provide instructions or data to the processing device. The software may also be distributed over network-coupled computer systems, and stored or executed in a distributed manner. The software and data may be stored on one or more computer-readable recording media.

[0085] The method according to the embodiment may be embodied in the form of program instructions executable by various computer means and recorded on a computer readable medium. The computer readable medium may include program instructions, data files, data structures, and the like, alone or in combination. The program instructions recorded on the medium may be specially designed and constructed for the embodiment, or may be known and available to those skilled in the art of computer software. Examples of computer readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes, optical media such as CD-ROMs and DVDs, magneto-optical media such as floptical disks, and hardware devices specially configured to store and execute program instructions, such as ROMs, RAMs, flash memories, and the like. Examples of program instructions include not only machine code, such as that produced by a compiler, but also high-level language code executable by a computer using an interpreter, and the like. The hardware devices may be configured to operate as one or more software modules to perform the operations of the present invention, and vice versa.

[0086] Although the embodiments have been described above based on limited examples and drawings, those skilled in the art may make various modifications and variations from the above disclosure. For example, the described techniques may be performed in a different order than described, and / or the components of the described systems, structures, devices, circuits, etc. may be combined or combined in a different manner than described, or may be counterbalanced or substituted with other components or equivalents to achieve suitable results.

[0087] Accordingly, other embodiments, examples, and equivalents of the following claims are intended to be within the scope of the following claims.

Claims

1. A method for predicting a neoantigen using a neoantigen determination apparatus, comprising: receiving as input a peptide sequence and an HLA allele sequence extracted from a target cancer tissue; measuring T cell activity data for the peptide, calculating T cell activity data for the peptide in the cancer tissue using a table or database in which the measured data is recorded, thereby obtaining T cell activity data from the peptide sequence, inputting the T cell activity data into an immune activity prediction model, and outputting a first predicted value for predicting the immune activity of the peptide sequence; measuring binding data related to binding strength for all binding relationships between peptides and HLA alleles, calculating binding data between the peptides and HLA alleles of the target cancer tissue using a table or database in which the measured binding data is recorded, thereby obtaining binding data from the HLA allele sequence, inputting the binding data into a binding prediction model, and outputting a second predicted value predicting binding between the peptide sequence and the HLA allele sequence; inputting the T cell activity data and the binding data into an immune resistance prediction model to output a third predicted value for predicting immune resistance of the target cancer tissue; a step of inputting the first to third predicted values ​​into a model trained with an output of N / Y corresponding to whether or not the target cell is a neoantigen, and outputting N / Y corresponding to whether or not the target cell is a neoantigen, thereby generating neoantigen information for the target cell using the first to third predicted values.

2. The method for predicting neoantigens using peptide sequences and HLA allele sequences according to claim 1, wherein at least one of the immune prediction model, the binding prediction model, and the immune resistance prediction model is trained by a machine learning algorithm based on a training dataset including peptide sequences and HLA allele sequences present in a plurality of target cancer tissues.

3. 3. The method of claim 2, wherein the target cancer tissue comprises cells engineered to express a single MHC class I or class II allele.

4. The method of predicting neoantigens using peptide sequences and HLA allele sequences according to claim 2, wherein the target cancer tissues include human cells obtained or derived from multiple patients. Law.

5. The method of claim 2, wherein the target cancer tissues include fresh or frozen tumor cells obtained from multiple patients.

6. The method of claim 2, wherein the target cancer tissues include fresh or frozen tissue cells obtained from multiple patients.

7. The method of claim 2, wherein the target cancer tissue comprises a peptide identified using a T cell assay.

8. The method for predicting neoantigens using peptide sequences and HLA allele sequences according to claim 2, wherein the training dataset includes at least one of data related to a protein sequence associated with the target cancer tissue, data related to an MHC peptide sequence associated with the target cancer tissue, binding data between a peptide associated with the target cancer tissue and an HLA allele, data related to a transcriptome associated with the target cancer tissue, and data related to a genome associated with the target cancer tissue.

9. The method for predicting neoantigens using peptide sequences and HLA allele sequences described in claim 1, wherein the immune prediction model is a model trained using T cell activity data from a peptide sequence as input and the immune activity of the peptide sequence as output.

10. The method for predicting neoantigens using peptide sequences and HLA allele sequences according to claim 1, wherein the binding prediction model is a model trained using binding data from HLA allele sequences and peptide sequences as input and binding properties of the peptide sequences and the HLA allele sequences as output.

11. The method for predicting neoantigens using peptide sequences and HLA allele sequences according to claim 1, wherein the immune tolerance prediction model is a model trained using T cell activity data from peptide sequences and HLA allele sequences and binding data from HLA allele sequences and peptide sequences as inputs, and immune tolerance between peptide sequences and HLA allele sequences as output.

12. A computer program stored on a computer-readable storage medium for causing a computer to carry out the method according to any one of claims 1 to 11.

Images