Method for identifying breast cancer prognosis fusion gene target and antigen candidate and application

Abstract

The invention provides a method for identifying a breast cancer prognosis fusion gene target and an antigen candidate and application. The method comprises the following steps: extracting a fusion gene existing in a breast cancer patient, evaluating the prognosis of the patient, and providing a panorama of the prognosis fusion gene in the breast cancer patient; identifying a fusion protein from the fusion gene, generating a full-length fusion protein sequence, and annotating a functional structure domain; extracting a fusion peptide from the fusion protein, predicting the combination of the fusion peptide and the HLA antigen, and forming an immunotherapy candidate library of the breast cancer neoantigen; the binding site of the breast cancer neoantigen and the MHC molecule is predicted, and the reliability of the breast cancer neoantigen is verified. The fusion protein is recognized from the fusion gene, the fusion peptide is extracted, the new antigen candidate of the BRCA is predicted through high-throughput artificial intelligence calculation, a new target spot is provided for cancer immunodetection, the new antigen candidate can be used as a potential predictive factor for tumor survival prognosis and immune checkpoint response, and a new way is provided for personalized medical treatment and immunodetection.

Description

Technical Field

[0001] The present invention relates to the field of biomedical technologies, and in particular, to the field of markers related to tumor prognosis and treatment; specifically, it relates to a method for identifying fusion gene targets and antigen candidates for breast cancer prognosis and its applications. Background Art

[0002] A fusion gene is formed by the fusion of fragments of two independent genes due to chromosomal translocation or deletion. Fusion genes can produce abnormal transcripts or proteins, which may initiate or promote tumorigenesis. The relationship between fusion genes and tumor phenotypes makes these genetic abnormalities crucial for tumor detection and subclassification, improves the accuracy of risk stratification, and provides opportunities for targeted therapy.

[0003] Tumor neoantigens are peptide sequences generated by genetic variations in the tumor genome. The binding of neoantigens to human leukocyte antigen (HLA) is crucial for antigen presentation, which is a necessary prerequisite for effective T cell recognition. Neoantigens in tumor therapy go beyond traditional therapeutic targets and minimize the risk of autoimmune reactions, increasing the diversity of neoantigens available for immunotherapy strategies (immune checkpoint inhibitors, personalized tumor vaccines, and adoptive T cell therapies). Traditional neoantigen research has mainly focused on neoantigens from single nucleotide variations (SNVs) and insertions or deletions (Indels).

[0004] Fusion genes generate new open reading frames (ORFs), producing peptides that are significantly different from self-antigens, providing a new approach for immunotherapy. Compared with SNVs and Indels, gene fusions can generate more neoantigens, and the immunogenicity of fusion-derived neoantigens is enhanced, indicating that gene fusions would be promising targets for the next generation of cancer immunotherapy. Gene fusion neoantigens are a new direction for tumor immunotherapy.

[0005] Breast cancer (BRCA) is the most commonly diagnosed malignant tumor among women globally and the fifth leading cause of cancer-related deaths. Although the progress of immunotherapy at the current stage provides promising treatment options for BRCA, there are still challenges such as limited efficacy, significant side effects, and drug resistance. Therefore, there is an urgent need to develop new BRCA tumor-specific fusion neoantigens. Summary of the Invention

[0006] In view of this, the first objective of the present invention is to design a method for identifying fusion gene targets and antigen candidates for breast cancer prognosis, identify fusion proteins from fusion genes and extract fusion peptides, accurately predict neoantigen candidates for breast cancer BRCA through high-throughput artificial intelligence calculations, provide new targets for personalized cancer immunodetection, and can be used as potential predictors of tumor survival prognosis and response to immune checkpoints.

[0007] The second object of the present invention is to provide the application of the method for identifying fusion gene targets and antigen candidates for breast cancer prognosis in tumor immune detection.

[0008] In order to achieve the above objects of the present invention, the following technical solutions are specifically adopted: The present invention provides a method for identifying fusion gene targets and antigen candidates for breast cancer prognosis, which identifies fusion proteins from fusion genes and extracts fusion peptides, and predicts new antigen candidates for breast cancer BRCA through high-throughput artificial intelligence calculation, including the following steps: Identifying fusion proteins from fusion genes and extracting fusion peptides, and accurately predicting new antigen candidates for breast cancer BRCA through high-throughput artificial intelligence calculation, including the following steps: Extracting the fusion genes existing in breast cancer patients, using the Deepsurv model to evaluate the prognosis of the patients, and providing a panorama of the prognostic fusion genes in breast cancer patients; Identifying fusion proteins from fusion genes, using the AGFusion tool to generate full-length fusion protein sequences, and annotating the functional domains of the full-length fusion protein sequences; Specifically, use AGFusion to predict the full-length complementary DNA (cDNA) sequence, the corresponding coding sequence (CDS), and the protein sequence, and then annotate them using the Pfam domain.

[0009] Extracting fusion peptides from fusion proteins, using a Transformer neural network and attention mechanism-based analysis to predict the binding of fusion peptides to HLA antigens (preferably 112 types), and forming an immunotherapy candidate library for breast cancer neoantigens (preferably, the number of data n = 7791, covering 88.41% of the patients), wherein 15 neoantigens are verified and actually converted into mass spectrometry data of breast cancer patients; Predicting the binding sites of breast cancer neoantigens to MHC (HLA) molecules, and verifying the reliability of the breast cancer neoantigens.

[0010] Specifically, the present invention identifies a prognostic neoantigen (QLVDFQWKL) from the TBC1D4--COMMD6 fusion gene, QLVDFQWKL significantly improves the prognosis of patients and binds extensively to 16 HLA alleles (Figure 4b), and these highly immunogenic and tumor-specific neoantigens provide new targets for personalized cancer immune detection and can be used as potential predictors of tumor survival prognosis and response to immune checkpoint therapy.

[0011] Furthermore, the method for generating a full-length fusion protein sequence using the AGFusion tool includes: determining the fusion proteins significantly associated with patient survival through a T-test, and examining the difference between the survival rates of patients with fusion proteins and those without fusion proteins. The calculation expression of the T-test is as follows:

[0012] In formula (1), n1 and n2 are the sizes of the two samples respectively, and n1 and n2 are the standard deviations of the two samples respectively;

[0013] In formula (2), and are the means of the two samples respectively, and t represents whether the difference between the averages of the two samples is significant; Take the first 20 significant prognostic fusion protein sequences as specific breast cancer prognostic markers.

[0014] Furthermore, the method for extracting fusion peptides from fusion proteins includes: Since the longest peptide binding to MHC molecules is 15 amino acids, at most 8 amino acids are extracted upstream of the fusion protein junction point and at most 8 amino acids are extracted downstream; Delete the sequences that match 100% in the Swissprot database (protein sequence database) to obtain a specific list of fusion peptides (preferably, n = 3198).

[0015] Furthermore, the method for predicting the binding of fusion peptides to 112 HLA antigens to form an immunotherapy candidate library for breast cancer neoantigens includes: Use the netMHCpan-4.1 artificial neural network to predict the binding of fusion peptide sequences to MHC class I molecules; extract the peptides strongly binding to MHC (HLA) class I molecules: p-MHCI, n = 7988, and filter through Rank_EL<0.5; further predict the reliable Rp-MHCI peptides through TransPHLA (based on Transformer) to form an immunotherapy candidate library for breast cancer neoantigens.

[0016] Furthermore, the method for predicting the binding of fusion peptides to HLA antigens includes: Download the mass spectrometry proteome data of breast cancer from the TCGA proteome dataset PDC000173, match the Rp-MHCI peptides with the mass spectrometry proteome data to obtain ms-Rp-MHCI peptides, and confirm that the predicted fusion genes containing the Rp-MHCI peptides are translated into proteins.

[0017] Further, the method for predicting the binding site of the predicted breast cancer neoantigen to the MHC molecule includes: Use netMHCpan and TransPHLA to predict the binding of the fusion peptide QLVDFQWKL to MHC class I molecules.

[0018] Further, the method for predicting the binding site of the predicted breast cancer neoantigen to the MHC molecule further includes: Based on the structures of HLA-A and HLA-B of AlphaFold2, construct a molecular dynamics model of HLA and the neoantigen QLVDFQWKL, and apply the AlphaFold2 model to predict the binding of the fusion peptide QLVDFQWKL to MHC class I molecules.

[0019] The present invention also provides the application of the method for identifying breast cancer prognosis fusion gene targets and antigen candidates as described above in tumor immune detection.

[0020] The present invention also provides an information data processing terminal, which includes a memory and a processor. The memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the method for identifying breast cancer prognosis fusion gene targets and antigen candidates as described above.

[0021] The present invention also provides a computer-readable storage medium, which stores a computer program, and when the computer program is executed by a processor, the processor executes the steps of the method for identifying breast cancer prognosis fusion gene targets and antigen candidates as described above.

[0022] Compared with the prior art, the beneficial effects of the present invention are: The method for identifying breast cancer prognosis fusion gene targets and antigen candidates provided by the present invention identifies fusion proteins from fusion genes and extracts fusion peptides, and accurately predicts neoantigen candidates of breast cancer BRCA through high-throughput artificial intelligence calculation, providing new targets for cancer immune detection, and can be used as potential predictors of tumor survival prognosis and response to immune checkpoints, providing new ways for personalized medicine and immune detection. Description of the Drawings

[0023] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings required to be used in the embodiments of the present application. Obviously, the following described drawings are only some embodiments of the present application, and those of ordinary skill in the art can also obtain other drawings based on these drawings without creative efforts.

[0024] Figure 1 It is a panoramic view of prognosis fusion genes in breast cancer patients provided by the embodiments of the present invention; Figure 2 This is the experimental schematic diagram of the fusion gene in the embodiment of the present invention generating a chimeric protein with new functions; Figure 3 This is the experimental schematic diagram of generating a potential new antigen library from the fusion peptide in the embodiment of the present invention; Figure 4 This is the experimental schematic diagram of the TBC1D4--COMMD6 fusion generating a prognostic neoantigen in the embodiment of the present invention. Detailed implementation manners

[0025] The following will describe the implementation schemes of the present invention in detail in conjunction with embodiments. However, those skilled in the art will understand that the following embodiments are only used to illustrate the present invention and should not be regarded as limiting the scope of the present invention. For those not specified in the embodiments, they are carried out according to conventional conditions or conditions recommended by the manufacturer. For reagents or instruments not specified by the manufacturer, they are all conventional products that can be obtained through commercial purchase. Embodiment

[0026] 1. Extract prognostic fusion genes in breast cancer patients: A fusion gene is formed by the fusion of two independent genes (as shown in part a of Figure 1 ), and it serves as an important diagnostic, prognostic, and therapeutic marker in tumors. In this embodiment, the fusion genes existing in breast cancer patients (n = 864) were extracted. There were a total of 5279 breast cancer fusion genes, and the maximum number of fusions in a single patient was 41 (as shown in part b of Figure 1 ).

[0027] Fusion genes are divided into intrachromosomal fusions (occurring on the same chromosome) and interchromosomal fusions (occurring between different chromosomes) (as shown in part c of Figure 1 ). Further analyze the chromosomal positions of the breakpoints of these fusion genes to show differences in chromosomal instability. Chromosome 17 showed the highest number of fusion events (n = 1319), followed by chromosome 1, with 955 fusion events (as shown in part d of Figure 1 ). There were 4286 intrachromosomal fusions and 417 interchromosomal fusions (as shown in part e of Figure 1 ). The highest number of intrachromosomal fusions occurred on chromosome 11 (n = 51), while the most common interchromosomal fusion occurred between chromosome 17 and chromosome 1 (n = 77) (as shown in part e of Figure 1 ). A total of 5953 genes were involved in the fusion events (partner genes). SHANK2 located on chromosome 11 was the most active gene, having 41 fusions with other genes, followed by FBXL20 and BCAS3 on chromosome 17 (as shown in part e of Figure 1as shown in part f).

[0028] Using DeepSurv to evaluate the prognosis of high-frequency partner genes involved in more than five fusions, there were 58 prognostic partner genes in this example. Among them, 34 partner genes were significantly associated with a significant increase in patient survival rate, while 24 partner genes were significantly associated with a significant decrease in survival rate. Nine patients with PPP1R1B fusion were identified (such as Figure 1 as shown in parts f and g), and the survival rate of patients carrying these fusions was significantly reduced, with the median survival time decreasing from 4267 days to 2911 days (as Figure 1 shown in part h).

[0029] 2. Fusion genes produce chimeric proteins with new functions: Using AGFusion to predict the full-length complementary DNA (cDNA) sequence, the corresponding coding sequence (CDS), and the protein sequence, and then annotating these sequences with Pfam domains (as Figure 2 shown in parts a, b, and c), it was determined that 3866 (73.13%) fusion genes were translated into proteins. (Such as the PPP1R1B--PIPOX and SIK3--PPP1R1B fusion proteins shown in parts b and c). The maximum protein length was 8544 amino acids (as Figure 2 shown in part d). Figure 2

[0030] Through the T-test, 74 fusion proteins were determined to be significantly associated with patient survival; examining the difference in the survival rates of patients with fusion proteins and patients without fusion proteins in two samples, the calculation expression of the T-test is: (1) In formula (1), n1 and n2 are the capacities of the two samples respectively, and n1 and n2 are the standard deviations of the two samples respectively;

[0031] In formula (2), and are the means of the two samples respectively, and t represents whether the difference in the means of the two samples is significant; Among them, the first 20 significant prognostic fusion proteins are listed in Figure 2 part e and can be used as specific breast cancer prognostic markers. Patients with the TDRD3-DIAPH3 fusion protein showed significantly better survival (median overall survival: 7455 days), while patients with the TNRC18--RNF216 fusion protein had the lowest survival rate (median overall survival: 2763 days) (as Figure 2 ​as shown in part f of). In addition, 3993 domains were identified in 3035 fusion proteins, and the top three were transmembrane helices (TMhelix, n = 2033), zinc finger C2H2-type domains (zf-C2H2, n = 215), and spectrin (n = 160) (as Figure 2 shown in part g of). The MAGI3-NOTCH2 fusion protein contains 41 domains, which is also the highest number in a single fusion protein.

[0032] Fusion genes combine protein domains in a new way to produce chimeric proteins, forming new functions that drive resistance to tumors.

[0033] 3. Generation of a potential new antigen library from fusion peptides: To ensure that the fusion peptides are new sequences generated from the patient's fusion genes, the following steps were taken (as shown in part a of Figure 3 and part b of Figure 3 ): 1) Peptide extraction: Since the longest peptide that binds to MHC molecules is 15 amino acids, at most 8 amino acids were extracted upstream and 8 amino acids were extracted downstream of the fusion protein junction; 2) Sequences that match 100% in the Swissprot database were deleted to obtain a list of unique fusion peptides (n = 3198). The binding of fusion peptide sequences to MHC class I molecules was predicted using netMHCpan-4.1 (predicted by an artificial neural network). Peptides that strongly bind to MHC (HLA) class I molecules (p-MHCI, n = 7988) were extracted (filtered by Rank_EL < 0.5) (as shown in part b of Figure 3 ). Further, reliable Rp-MHCI peptides (n = 7971) were predicted by TransPHLA (based on Transformer) to form a new antigen library for breast cancer (as shown in part b of Figure 3 ). The distribution of potential new antigens in breast cancer patients was analyzed to evaluate the treatment scope, and 88.41% of breast cancer patients had Rp-MHCI peptides.

[0034] In addition, mass spectrometry proteomic data of breast cancer were downloaded from the TCGA proteome dataset (PDC000173) to confirm that the predicted fusion genes containing Rp-MHCI peptides were translated into proteins. There were 23 Rp-MHCI peptides in the patient proteome dataset that matched the mass spectrometry data (ms-Rp-MHCI peptides), and these ms-Rp-MHCI peptides came from 15 fusion proteins, and their sequences and electrostatic structures are as shown in Figure 3As shown in part c. The electrostatic structure predicted by AlphaFold2 shows the three-dimensional conformation of the peptides, revealing their spatial arrangement and folding pattern at the molecular level, providing a structural basis for in vitro synthesis. Neoantigens derived from fusion genes have great potential in breast cancer immunotherapy. The Rp-MHCI peptides identified in this example, especially the ms-Rp-MHCI peptides that match the mass spectrometry data, provide a new approach for personalized cancer immunodetection.

[0035] 4. Validation of prognostic neoantigens generated by TBC1D4--COMMD6 fusion: Validate the reliability of the prognostic neoantigen from the TBC1D4–COMMD6 fusion protein (as shown in part a of Figure 4 ), which increased the median survival time of patients to 6456 days. Figure 4 Parts b-d of Figure 4 show the binding of the fusion peptide QLVDFQWKL to MHC class I molecules, as predicted by netMHCpan and TransPHLA. The presence of this neoantigen peptide was further verified by matching the peak spectra in the mass spectrometry data (as shown in part e of Figure 4 ). Based on the structures of HLA-A and HLA-B from AlphaFold2, a molecular dynamics model of HLA and the neoantigen QLVDFQWKL was constructed, and the results showed a very strong binding force, which was consistent with the predictions of netMHCpan and TransPHLA (as shown in part f of Figure 4 ). Multiple validation results enhance the feasibility of the neoantigen as a candidate for breast cancer immunotherapy, thus facilitating the design of potential treatment regimens, such as using cell surface fusion proteins as targets for chimeric antigen receptor T (CAR-T) cells (as shown in part g of

[0036] Neoantigens derived from fusion genes have great potential in immunotherapy, facilitating the development of personalized cancer vaccines and advancing precision medicine in oncology. The significant correlation between the 74 fusion proteins and patient survival in this example indicates that these biomarkers can be stratified according to the patient's prognosis. For example, the TDRD3-DIAPH3 fusion protein is associated with better survival outcomes, while the TNRC18-RNF216 fusion protein is associated with a poorer prognosis. Generating full-length complementary DNA (cDNA), coding sequences (CDS), and protein sequences using AGFusion and annotating them with Pfam domains provides a comprehensive view of the functional impact of these fusion genes, identifying the most active fusion partners, such as SHANK2, FBXL20, and BCAS3, which can be potential targets for intervening in the prognosis.

[0037] Neoantigens derived from gene fusions have broad application prospects for immunoassays. Neoantigens are tumor-specific and provide unique targets for immune system activation. In this example, advanced prediction tools such as netMHCpan and TransPHLA were used to identify and validate neoantigen peptides that can bind to MHC class I molecules. Subsequent validation with mass spectrometry data confirmed the presence of these neoantigens, making them viable candidates for vaccine development. For example: the neoantigen (QLVDFQWKL) derived from the TBC1D4–COMMD6 fusion protein, which significantly increased the median survival of patients (by 2189 days, as shown in part a of Figure 4 ). This indicates that targeting neoantigens derived from fusion genes can improve the treatment outcomes of patients, and this approach can be extended to other cancers with similar genomic alterations. Although the invention has been illustrated and described with reference to specific embodiments, it will be appreciated that many other changes and modifications can be made without departing from the spirit and scope of the invention. Accordingly, this means that all such changes and modifications that fall within the scope of the present invention are included in the appended claims.

Claims

1. A method for identifying breast cancer prognosis fusion gene targets and antigen candidates, characterized in that: Identify fusion proteins and extract fusion peptides from fusion genes, and predict new antigen candidates for breast cancer BRCA through high-throughput artificial intelligence calculations, including the following steps: Extract fusion genes present in breast cancer patients, use Deepsurv model to evaluate the prognosis of patients, and provide a panoramic view of prognostic fusion genes in breast cancer patients; Identify fusion proteins from fusion genes, generate full-length fusion protein sequences using the AGFusion tool, and annotate the functional domains of the full-length fusion protein sequences; Extract fusion peptides from fusion proteins, use Transformer neural networks and attention-based analysis to predict the binding of fusion peptides to HLA antigens, and form a candidate library for immunotherapy of breast cancer neoantigens; Predict the binding sites of breast cancer neoantigens and MHC molecules and verify the reliability of the breast cancer neoantigens.

2. The method for identifying breast cancer prognosis fusion gene targets and antigen candidates according to claim 1, characterized in that: The method for generating a full-length fusion protein sequence using the AGFusion tool includes: determining the fusion protein significantly associated with patient survival by a T test, and examining the difference between the survival rate of patients with the fusion protein and the survival rate of patients without the fusion protein. The calculation expression of the T test is: （1） In formula (1), n1 and n2 are the capacity of the two samples, and n1 and n2 are the standard deviations of the two samples; （2） In formula (2), , are the means of the two samples respectively, and t indicates whether the difference between the means of the two samples is significant; The top 20 significant prognostic fusion protein sequences were used as specific breast cancer prognostic markers.

3. The method for identifying breast cancer prognosis fusion gene targets and antigen candidates according to claim 1, characterized in that: The method for extracting fusion peptide from fusion protein comprises: A maximum of 8 amino acids were extracted upstream of the fusion protein junction and a maximum of 8 amino acids were extracted downstream; Sequences with 100% matches in the Swissprot database were deleted to obtain a specific list of fusion peptides.

4. The method for identifying breast cancer prognosis fusion gene targets and antigen candidates according to claim 1, characterized in that: The method for predicting the binding of fusion peptides to HLA antigens to form a candidate library for immunotherapy of breast cancer neoantigens comprises: The netMHCpan-4.1 artificial neural network was used to predict the binding of fusion peptide sequences to MHC class I molecules; peptides that strongly bind to MHC class I molecules were extracted: p-MHCI, filtered by Rank_EL < 0.5; and reliable Rp-MHCI peptides were further predicted by TransPHLA to form a candidate library for immunotherapy of breast cancer neoantigens.

5. The method for identifying breast cancer prognosis fusion gene targets and antigen candidates according to claim 4, characterized in that: The method for predicting the binding of a fusion peptide to an HLA antigen comprises: Mass spectrometry proteomic data of breast cancer were downloaded from the TCGA proteomic data set PDC000173, and the Rp-MHCI peptide was matched with the mass spectrometry proteomic data to obtain the ms-Rp-MHCI peptide, and it was confirmed that the predicted fusion gene containing the Rp-MHCI peptide was translated into protein.

6. The method for identifying breast cancer prognosis fusion gene targets and antigen candidates according to claim 1, characterized in that: The method for predicting the binding site of breast cancer neoantigens and MHC molecules comprises: netMHCpan and TransPHLA were used to predict the binding of the fusion peptide QLVDFQWKL to MHC class I molecules.

7. The method for identifying breast cancer prognosis fusion gene targets and antigen candidates according to claim 1, characterized in that: The method for predicting the binding site between breast cancer neoantigen and MHC molecule also includes: Based on the structures of HLA-A and HLA-B of AlphaFold2, a molecular dynamics model of HLA and the neoantigen QLVDFQWKL was constructed, and the AlphaFold2 model was used to predict the binding of the fusion peptide QLVDFQWKL to MHC class I molecules.

8. Use of the method for identifying breast cancer prognosis fusion gene targets and antigen candidates according to any one of claims 1 to 7 in tumor immunoassay.

9. An information data processing terminal, characterized in that: The information data processing terminal includes a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the method for identifying breast cancer prognosis fusion gene targets and antigen candidates according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: A computer program is stored, and when the computer program is executed by a processor, the processor executes the steps of the method for identifying breast cancer prognosis fusion gene targets and antigen candidates according to any one of claims 1 to 7.

Images