High-throughput detection platform for thyroid cancer specific fusion gene CCDC6

Abstract

The invention relates to the technical field of fusion gene detection, and discloses a high-throughput detection platform of a thyroid cancer specific fusion gene CCDC6. According to the detection platform disclosed by the invention, high-throughput and high-sensitivity detection on the thyroid cancer CCDC6 fusion gene is realized through a technical scheme of combining sample enrichment and deep learning recognition. Through combination of the sample enrichment module and the deep learning recognition module, the sample enrichment module promotes high-sensitivity recognition of the deep learning recognition module, an end-to-end integrated solution is provided, a large number of samples can be processed in one-time operation, the result can be automatically interpreted, and the detection efficiency and accuracy of the CCDC6 fusion gene are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of fusion gene detection, and in particular to a high-throughput detection platform for the thyroid cancer-specific fusion gene CCDC6. Background Art

[0002] Thyroid cancer is the most common malignant tumor of the endocrine system, among which papillary thyroid carcinoma (PTC) accounts for more than 80% of all thyroid cancer cases. Studies have shown that multiple gene fusion events play a key driving role in the occurrence and development of thyroid cancer, especially the rearrangement and fusion of the RET gene, such as RET-CCDC6 (RET / PTC1) and RET-NCOA4 (RET / PTC3). These two fusion types account for about 90% of RET fusion cases in PTC. Among them, The RET-CCDC6 fusion gene can abnormally activate oncogenic signaling pathways such as MAPK and PI3K-AKT, promoting tumorigenesis. Therefore, its detection is of great significance for the accurate diagnosis, prognosis assessment, and targeted treatment selection of thyroid cancer. For example, the selective RET inhibitor Selpercatinib, approved by the US FDA in 2024, is used for patients with RET fusion-positive thyroid cancer. Only when RET fusion patients are accurately identified can they benefit from this type of targeted therapy.

[0003] Currently, commonly used methods for fusion gene detection include reverse transcription-PCR (RT-PCR) or real-time quantitative PCR (qPCR), and fluorescence in situ hybridization (FISH). RT-PCR or qPCR uses primers and probes designed specifically for known fusion transcripts to detect fusion genes through amplification. Traditional RT-PCR or qPCR methods have the following disadvantages: they can only detect one fusion type at a time, resulting in low throughput; they are prone to missing fusions in low-abundance samples (such as circulating tumor DNA (ctDNA) in plasma). FISH uses fluorescently labeled probes to visualize gene translocations in tissue sections. However, FISH has the following disadvantages: it is cumbersome to perform, relies on microscopic interpretation, and therefore cannot achieve high-throughput detection. Furthermore, it has difficulty detecting unknown fusion types.

[0004] In summary, existing fusion gene detection methods have the following problems: 1. Low throughput: Most methods require testing samples one by one, which makes it difficult to meet the needs of large-scale screening.

[0005] 2. High dependence on manual labor: The interpretation of results relies on experience and is prone to subjective errors, such as microscopic observation of FISH or threshold setting of PCR.

[0006] 3. Lack of integrated solutions: From sample preparation, target enrichment to data analysis, multiple platforms need to be switched, the process is cumbersome and the sample loss is large.

[0007] Therefore, there is an urgent need to develop a highly sensitive, high-throughput, and automated fusion gene detection method to meet people's needs. Summary of the Invention

[0008] In order to solve the above technical problems, the present invention provides a high-throughput detection platform for thyroid cancer-specific fusion gene CCDC6 (RET-CCDC6).

[0009] The specific technical solutions of the present invention are: In a first aspect, the present invention provides a high-throughput detection platform for the thyroid cancer-specific fusion gene CCDC6, comprising: A sample enrichment module is used to selectively capture and enrich the CCDC6 fusion gene sequence in the sample to be tested; Deep learning recognition module, which includes a pre-trained deep learning model for automatic identification of fusion genes in enriched sequence data; The result output and verification module is used to generate test reports and support result verification.

[0010] The detection platform of this invention achieves high-throughput, highly sensitive detection of the CCDC6 fusion gene in thyroid cancer through a "sample enrichment + deep learning recognition" technical solution. By combining a sample enrichment module with a deep learning recognition module, the sample enrichment module facilitates the high-sensitivity recognition of the deep learning recognition module, providing an end-to-end integrated solution. This solution can process a large number of samples and automatically interpret the results in a single run, significantly improving the efficiency and accuracy of CCDC6 fusion gene detection.

[0011] As a preferred embodiment of the above detection platform, the sample enrichment module is captured by at least one of the following methods: Designing a biotin-labeled probe system targeting the fusion breakpoint region of the RET gene and the CCDC6 gene, and then capturing the fusion gene through the biotin-labeled probe system; A specific primer amplification system is designed, and the fusion gene is captured by the specific primer amplification system, wherein the sequence of the specific primer covers the RET-CCDC6 fusion junction sequence.

[0012] Biotinylated probes are designed to cover the breakpoints of the RET and CCDC6 gene fusions. After hybridization, fragments containing the fusion sequence are enriched using magnetic beads, or pre-amplified using specific primers that cover the RET-CCDC6 fusion junction. This enrichment process can improve the detection rate of low-abundance fusion genes in subsequent analysis.

[0013] The detection platform provided by the present invention adopts a biotin-labeled probe system targeting the fusion breakpoint region of the RET gene and the CCDC6 gene, and a specific primer amplification system covering the RET-CCDC6 fusion junction sequence. It can accurately capture low-abundance fusion genes, such as 0.1% RET-CCDC6 in plasma, and has the characteristics of high sensitivity.

[0014] More preferably, the biotin-labeled probe system comprises an oligonucleotide probe pool covering exons 11-12 of the RET gene and exons 1-2 of the CCDC6 gene, and the probe length is within the range of 50-100 bp.

[0015] As a preferred embodiment of the above-mentioned detection platform, the deep learning recognition module adopts a convolutional neural network (CNN) or Transformer architecture, and the input data includes at least one of the sequencing read sequence, breakpoint position characteristics and supporting read number.

[0016] Further preferably, the deep learning model is trained by the following steps: Input sequencing data of known RET-CCDC6 fusion positive and negative samples; A one-dimensional convolutional layer is used to extract sequence features, and an LSTM network is used to analyze cross-site dependencies. Output the probability score of the fusion gene existence and monitor it through the validation set to prevent overfitting.

[0017] As a preferred embodiment of the above detection platform, the sample to be tested is plasma circulating tumor DNA.

[0018] The platform's detection sensitivity reaches 0.1% mutant allele frequency, making it suitable for monitoring minimal residual disease in plasma circulating tumor DNA (ctDNA).

[0019] As a preferred embodiment of the above detection platform, the result output and verification module further includes: The function of automatically guiding positive samples to digital PCR (ddPCR) verification; Standardized report generation interface for interfacing with laboratory information management systems.

[0020] In a second aspect, the present invention provides a method for detecting thyroid cancer fusion genes, using the above-mentioned detection platform, which comprises the steps of: Extract nucleic acids from tissue or plasma samples and enrich for target sequences; Perform high-throughput sequencing or amplification signal collection on the enriched products; A deep learning model was used to analyze the data and determine the presence of the CCDC6 fusion gene; Output quantitative test results and verification suggestions.

[0021] As a preferred embodiment of the above method, no less than 96 samples are processed simultaneously in the target sequence enrichment step, and the single run time does not exceed 24 hours.

[0022] In a third aspect, the present invention provides a computer-readable storage medium storing a program for implementing the deep learning recognition module in the above-mentioned detection platform, wherein the program, when executed by a processor, can: receiving sequence data generated by sequencing or amplification; Call the pre-trained model to calculate the probability of the fusion gene; The output is a structured result containing the breakpoint locations and confidence scores.

[0023] Compared with the prior art, the present invention has the following technical effects: (1) The detection platform provided by the present invention adopts a biotin-labeled probe system targeting the breakpoint region of the RET gene and the CCDC6 gene, and a specific primer amplification system covering the RET-CCDC6 fusion junction sequence. It can accurately capture low-abundance fusion genes, such as 0.1% RET-CCDC6 in plasma, and has the characteristics of high sensitivity.

[0024] (2) The detection platform of the present invention automatically identifies fusion genes in enriched sequence data through a pre-trained deep learning model, and intelligent data analysis helps reduce the false positive / false negative rate of fusion gene detection.

[0025] (3) The detection platform of the present invention realizes end-to-end integrated detection and can perform high-throughput testing, effectively improving detection efficiency. DETAILED DESCRIPTION

[0026] The present invention will be further described below with reference to the following embodiments. Those skilled in the art will be able to implement the present invention based on these descriptions. Furthermore, the embodiments of the present invention described below are generally only a portion of the embodiments of the present invention, rather than all of the embodiments. Therefore, all other embodiments derived by those skilled in the art based on the embodiments of the present invention without inventive effort should fall within the scope of protection of the present invention.

[0027] Example 1 Provides a high-throughput detection platform for the thyroid cancer-specific fusion gene CCDC6 (RET-CCDC6), which includes: A sample enrichment module is used to selectively capture and enrich the CCDC6 fusion gene sequence in the sample to be tested; Deep learning recognition module, which includes a pre-trained deep learning model for automatic identification of fusion genes in enriched sequence data; The result output and verification module is used to generate test reports and support result verification.

[0028] The detection platform provided in this example achieves high-throughput, highly sensitive detection of the CCDC6 fusion gene in thyroid cancer through a "sample enrichment + deep learning recognition" technical solution. By combining a sample enrichment module with a deep learning recognition module, the present invention facilitates the high-sensitivity recognition of the deep learning recognition module, providing an end-to-end integrated solution capable of processing a large number of samples and automatically interpreting the results in a single run, significantly improving the efficiency and accuracy of CCDC6 fusion gene detection.

[0029] As a preferred embodiment of this invention, the sample enrichment module adopts at least one of the following methods for capture: Designing a biotin-labeled probe system targeting the fusion breakpoint region of the RET gene and the CCDC6 gene, and then capturing the fusion gene through the biotin-labeled probe system; A specific primer amplification system is designed, and the fusion gene is captured by the specific primer amplification system, wherein the sequence of the specific primer covers the RET-CCDC6 fusion junction sequence.

[0030] Biotinylated probes are designed to cover the breakpoints of the RET and CCDC6 gene fusions. After hybridization, fragments containing the fusion sequence are enriched using magnetic beads, or pre-amplified using specific primers that cover the RET-CCDC6 fusion junction. This enrichment process can improve the detection rate of low-abundance fusion genes in subsequent analysis.

[0031] The detection platform provided in this embodiment uses a biotin-labeled probe system targeting the breakpoint region of the RET gene and the CCDC6 gene, and a specific primer amplification system covering the RET-CCDC6 fusion junction sequence. It can accurately capture low-abundance fusion genes, such as 0.1% RET-CCDC6 in plasma, and has the characteristics of high sensitivity.

[0032] More preferably, the biotin-labeled probe system comprises an oligonucleotide probe pool covering exons 11-12 of the RET gene and exons 1-2 of the CCDC6 gene, and the probe length is within the range of 50-100 bp.

[0033] As a preference of this embodiment, the deep learning recognition module adopts a convolutional neural network (CNN) or a Transformer architecture, and the input data includes at least one of sequencing read sequences, breakpoint position features, and supporting read numbers.

[0034] Further preferably, the deep learning model is trained by the following steps: Input sequencing data of known RET-CCDC6 fusion positive and negative samples; A one-dimensional convolutional layer is used to extract sequence features, and an LSTM network is used to analyze cross-site dependencies. Output the probability score of the fusion gene existence and monitor it through the validation set to prevent overfitting.

[0035] As a preference in this embodiment, the sample to be tested is plasma circulating tumor DNA.

[0036] The platform's detection sensitivity reaches 0.1% mutant allele frequency, making it suitable for monitoring minimal residual disease in plasma circulating tumor DNA (ctDNA).

[0037] As a preferred embodiment of this invention, the result output and verification module further includes: The function of automatically guiding positive samples to digital PCR (ddPCR) verification; Standardized report generation interface for interfacing with laboratory information management systems.

[0038] Compared with the prior art, this embodiment has the following technical effects: (1) The detection platform provided by the present invention adopts a biotin-labeled probe system targeting the breakpoint region of the RET gene and the CCDC6 gene, and a specific primer amplification system covering the RET-CCDC6 fusion junction sequence. It can accurately capture low-abundance fusion genes, such as 0.1% RET-CCDC6 in plasma, and has the characteristics of high sensitivity.

[0039] (2) The detection platform of the present invention automatically identifies fusion genes in enriched sequence data through a pre-trained deep learning model, and intelligent data analysis helps reduce the false positive / false negative rate of fusion gene detection.

[0040] (3) The detection platform of the present invention realizes end-to-end integrated detection and can perform high-throughput testing, effectively improving detection efficiency.

[0041] Example 2 A method for detecting thyroid cancer fusion genes is provided, using the detection platform provided in Example 1, comprising the steps of: (1) Extract nucleic acids from tissue or plasma samples and enrich target sequences, specifically: A biotin-labeled probe covering the fusion breakpoint region of the RET gene and the CCDC6 gene is designed. After hybridization, magnetic beads are used to enrich fragments containing the fusion sequence, or specific primers are used for pre-amplification to enrich the fusion fragments, wherein the sequence of the specific primer covers the RET-CCDC6 fusion junction sequence, and the biotin-labeled probe system includes an oligonucleotide probe pool covering exons 11-12 of the RET gene and exons 1-2 of the CCDC6 gene, and the probe length is within the range of 50-100 bp.

[0042] The above enrichment process can improve the detection rate of low-abundance fusion genes in subsequent analysis.

[0043] (2) Perform high-throughput sequencing or amplification signal collection on the enriched products.

[0044] (3) Analyze the data and determine the presence of the CCDC6 fusion gene using a deep learning model. The deep learning model uses a convolutional neural network (CNN) and is trained by the following steps: input sequencing data of known RET-CCDC6 fusion-positive and -negative samples; extract sequence features using a one-dimensional convolutional layer and analyze cross-site dependencies using an LSTM network; output a probability score for the presence of the fusion gene, and monitor the validation set to prevent overfitting. Input data, including sequencing read sequences, breakpoint position features, and supporting read counts, to the trained deep learning model to output the presence of the CCDC6 fusion gene.

[0045] This release achieves high-throughput, high-sensitivity detection of the CCDC6 fusion gene in thyroid cancer through the combination of "sample enrichment + deep learning recognition". The target sequence enrichment step can process no fewer than 96 samples simultaneously, and the single run time does not exceed 24 hours.

[0046] For other technical effects of this embodiment, please refer to Example 1.

[0047] Example 3 A computer-readable storage medium is provided, storing a program for implementing the deep learning recognition module in the detection platform provided in Example 1, wherein the program, when executed by a processor, can: receiving sequence data generated by sequencing or amplification; Call the pre-trained model to calculate the probability of the fusion gene; The output is a structured result containing the breakpoint locations and confidence scores.

[0048] Unless otherwise specified, the raw materials and equipment used in the present invention are commonly used in the art; the methods used in the present invention are conventional methods in the art unless otherwise specified.

[0049] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any way. Any simple modification, change and equivalent transformation made to the above embodiment based on the technical essence of the present invention still fall within the scope of protection of the technical solution of the present invention.

Claims

1. A high-throughput detection platform for the thyroid cancer-specific fusion gene CCDC6, characterized by: include: A sample enrichment module is used to selectively capture and enrich the CCDC6 fusion gene sequence in the sample to be tested; Deep learning recognition module, which includes a pre-trained deep learning model for automatic identification of fusion genes in enriched sequence data; The result output and verification module is used to generate test reports and support result verification.

2. The high-throughput detection platform according to claim 1, wherein: The sample enrichment module is captured by at least one of the following methods: Designing a biotin-labeled probe system targeting the fusion breakpoint region of the RET gene and the CCDC6 gene, and then capturing the fusion gene through the biotin-labeled probe system; A specific primer amplification system is designed, and the fusion gene is captured by the specific primer amplification system, wherein the sequence of the specific primer covers the RET-CCDC6 fusion junction sequence.

3. The high-throughput detection platform according to claim 2, wherein: The biotin-labeled probe system comprises an oligonucleotide probe pool covering exons 11-12 of the RET gene and exons 1-2 of the CCDC6 gene, and the probe length is within the range of 50-100 bp.

4. The high-throughput detection platform according to claim 1, wherein: The deep learning recognition module adopts a convolutional neural network or a Transformer architecture, and the input data includes at least one of the sequencing read sequence, the breakpoint position characteristics and the number of supporting reads.

5. The high-throughput detection platform according to claim 4, wherein: The deep learning model is trained through the following steps: Input sequencing data of known RET-CCDC6 fusion positive and negative samples; A one-dimensional convolutional layer is used to extract sequence features, and an LSTM network is used to analyze cross-site dependencies. Output the probability score of the fusion gene existence and monitor it through the validation set to prevent overfitting.

6. The high-throughput detection platform according to claim 1, wherein: The sample to be tested is plasma circulating tumor DNA.

7. A method for detecting thyroid cancer fusion genes, using the detection platform according to any one of claims 1 to 6, characterized in that: Including steps: Extract nucleic acids from tissue or plasma samples and enrich for target sequences; Perform high-throughput sequencing or amplification signal collection on the enriched products; The data was analyzed using a deep learning model to determine the presence of the CCDC6 fusion gene.

8. The method according to claim 7, wherein: In the target sequence enrichment step, no fewer than 96 samples are processed simultaneously, and a single run time does not exceed 24 hours.

9. A computer-readable storage medium storing a program for implementing the deep learning recognition module in the detection platform according to any one of claims 1 to 6, characterized in that: When the program is executed by a processor, it is capable of: receiving sequence data generated by sequencing or amplification; Call the pre-trained model to calculate the probability of the fusion gene; The output is a structured result containing the breakpoint locations and confidence scores.