Auxiliary diagnosis system for benign and malignant thyroid nodules based on the combination of genomic variation and abnormal expression

By combining genomic mutation and abnormal expression, the problem of uncertainty in the diagnosis of thyroid nodules is solved, the diagnostic accuracy and reliability of treatment decisions are improved, and unnecessary treatment risks are reduced.

CN118248319BActive Publication Date: 2024-12-03SHANGHAI RIGEN BIOTECHNOLOGY CO LTD
View PDF 5 Cites 0 Cited by

Patent Information

Application Number
CN202410188901.1
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-02-20
Publication Date
2024-12-03
Estimated Expiration
2044-02-20

AI Technical Summary

Technical Problem

The prior art has diagnostic uncertainty in the diagnosis of benign and malignant thyroid nodules, with about 15% to 25% of cases not being diagnosed by fine needle aspiration cytology, resulting in clinical processing difficulties and unnecessary overdiagnosis and treatment.

Method used

Using an auxiliary diagnostic system based on the combination of genomic mutation and abnormal expression, the genomic data of the sample is obtained through the DNA data module and the RNA data module, and combined with biological information analysis and prediction models, a quantitative benign and malignant analysis method is provided.

Benefits of technology

It improves the accuracy of benign and malignant diagnosis of thyroid nodules, provides a more appropriate treatment and follow-up plan, reduces unnecessary diagnosis and treatment risks, and enhances the ability to evaluate the nature and pathological types of thyroid masses.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118248319B_ABST
    Figure CN118248319B_ABST
Patent Text Reader

Abstract

The present application provides a thyroid nodule benign and malignant auxiliary diagnosis system based on the combination of genome variation and abnormal expression, including: DNA data module: used to obtain the data of the sequencing library constructed by the shingled multiple amplification of the DNA of each sample; RNA data module: used to obtain the sequencing library data constructed by the multiple amplification after reverse transcription of the RNA of each sample and the RNA expression level data of the target genome of each sample: the first analysis module: used to perform bioinformatics analysis on the sequencing library data obtained in the DNA acquisition module and the RNA acquisition module, and obtain the benign and malignant judgment results of the thyroid nodules of each sample in combination with the somatic cell variation interpretation guide; the second analysis module: used to process and analyze the RNA expression level data of the target gene combination of each sample, and predict the benign and malignant of the thyroid nodules of each sample based on the constructed benign and malignant prediction model of the thyroid nodules to obtain the benign and malignant prediction results of the thyroid nodules of the corresponding samples; judgment module: used to judge the final benign and malignant judgment results of the thyroid nodules of each sample based on the benign and malignant judgment results of the thyroid nodules of each sample and the benign and malignant prediction results of the thyroid nodules of each sample. The present invention has important application value.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of molecular diagnosis, and in particular to an auxiliary diagnosis system for benign and malignant thyroid nodules based on the combination of genome variation and abnormal expression. Background Art

[0002] Thyroid cancer is a common malignant tumor in the endocrine system. Its typical manifestations include painless thyroid lumps, which may be caused by excessive proliferation of thyroid cells or cysts containing fluid. According to pathological classification, thyroid cancer can be divided into four main types: papillary, follicular, undifferentiated and medullary. Papillary and follicular carcinomas are classified as differentiated thyroid cancers, which together account for more than 90% of the total number of thyroid cancers. Although fine needle aspiration cytology is regarded as the standard procedure for determining whether thyroid lumps are benign or malignant, due to the overlapping cytological characteristics of benign and malignant lumps, approximately 15% to 25% of cases cannot be clearly diagnosed by this method, making the clinical treatment of these cases more difficult and sometimes leading to unnecessary overdiagnosis and treatment.

[0003] When the diagnosis cannot be confirmed by conventional methods, authoritative guidelines such as NCCN, ATA, and ACE recommend the use of molecular marker detection as an auxiliary means. Domestic guidelines and expert consensus also support the combination of fine needle aspiration and molecular marker detection to improve the accuracy of preoperative diagnosis. In this context, gene expression profiling plays an increasingly critical role in the auxiliary diagnosis of thyroid cancer. It can not only improve the accuracy of diagnosis, but also help to formulate more appropriate treatment and follow-up plans, thereby avoiding unnecessary diagnosis and treatment risks. Genetic testing is extremely valuable in distinguishing the nature of thyroid nodules, different pathological types of thyroid cancer, and even in the assessment of prognosis and the formulation of targeted treatment strategies. In order to further improve the performance of diagnostic tools, it is necessary not only to introduce newly discovered thyroid cancer-related molecular markers, but also to continuously evaluate and improve the technology for detecting new markers, such as driver mutations or gene fusion events within the genome. Summary of the invention

[0004] In view of the shortcomings of the prior art described above, the purpose of the present invention is to provide an auxiliary diagnosis system for benign and malignant thyroid nodules based on the combination of genomic variation and abnormal expression, so as to solve the problems of the prior art. On the basis of genomic variation detection, the present invention combines the detection of nodule tissue transcriptional expression markers, and provides a quantitative, highly scalable, and quantifiable benign and malignant analysis method, which has important application value in clinical practice.

[0005] To achieve the above-mentioned purpose and other related purposes, the present application provides, in a first aspect, a system for auxiliary diagnosis of benign and malignant thyroid nodules based on the combination of genomic variation and abnormal expression, comprising the following modules:

[0006] DNA data module: used to obtain the data of the sequencing library constructed by shingled multiple amplification of the DNA of each sample;

[0007] RNA data module: used to obtain the data of the sequencing library constructed by multiple amplification after reverse transcription of RNA of each sample and the RA expression level data of the target genome of each sample: wherein the target genome includes: 14 characteristic genes and 7 internal reference genes; the 14 characteristic genes are AGGF1, BANP, CDK12, CITED1, DICER1, ERBB3, ERBB4, FGFR2, FGFR4, IGF2BP3, KIT, MET, VHL, ZBTB8A;

[0008] The first analysis module is used to perform bioinformatics analysis on the data of the sequencing library obtained in the DNA data module and the RA data module, and obtain the benign or malignant judgment results of the thyroid nodules of each sample in combination with the somatic cell variation interpretation guide;

[0009] The second analysis module is used to process and analyze the RNA expression level data of the target genome of each sample, and to predict the benign and malignant nature of the thyroid nodules for each sample based on the constructed benign and malignant thyroid nodule prediction model to obtain the benign and malignant thyroid nodule prediction results of the corresponding samples;

[0010] Judgment module: used to judge the final benign or malignant judgment result of each sample based on the benign or malignant judgment result of each sample and the benign or malignant prediction result of each sample.

[0011] A second aspect of the present application provides a method for auxiliary diagnosis of benign and malignant thyroid nodules for non-disease diagnosis purposes, comprising the following steps:

[0012] 1) Obtaining data of sequencing libraries constructed by shingled multiplex amplification of DNA from each sample;

[0013] 2) Obtaining the sequencing library data constructed by multiple amplification after reverse transcription of RNA of each sample and the RA expression level data of the target genome of each sample; wherein the target genome includes: 14 characteristic genes and 7 internal reference genes; the 14 characteristic genes are AGGF1, BANP, CDK12, CITED1, DICER1, ERBB3, ERBB4, FGFR2, FGFR4, IGF2BP3, KT, MET, VHL, ZBTB8A;

[0014] 3) Performing bioinformatics analysis on the sequencing library data of step 1) and step 2), and combining the somatic cell variation interpretation guide to obtain the benign or malignant judgment result of the thyroid nodule;

[0015] 4) Processing and analyzing the RNA expression level data of the target genome in step 2), and obtaining a prediction result of benign and malignant thyroid nodules based on the constructed prediction model of benign and malignant thyroid nodules;

[0016] 5) Based on the benign or malignant judgment result of the thyroid nodule in step 3) and the benign or malignant prediction result of the thyroid nodule in step 4), a final benign or malignant judgment result of the thyroid nodule is obtained.

[0017] A third aspect of the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the analysis steps of step 3) to step 5) in the aforementioned method.

[0018] A fourth aspect of the present application provides a terminal, characterized by comprising: a processor and a memory;

[0019] The memory is used to store computer programs;

[0020] The processor is used to execute the computer program stored in the memory, so that the device performs the analysis steps of step 3) to step 5) in the aforementioned method.

[0021] Compared with the prior art, this application has the following beneficial effects:

[0022] Based on the detection of genomic variation, the present invention combines the detection of nodule tissue transcriptional expression markers and provides a set of quantitative, highly scalable and quantifiable benign and malignant analysis methods, which has important application value in clinical practice. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 This is a schematic diagram of the structure of a system for auxiliary diagnosis of benign and malignant thyroid nodules based on the combination of genomic variation and abnormal expression in one embodiment of the present invention.

[0024] Figure 2 The figure is a schematic diagram of the construction process of a model for predicting benign and malignant thyroid nodules in one embodiment of the present invention.

[0025] Figure 3 It is a schematic diagram of utilizing test data and evaluating model performance in one embodiment of the present invention.

[0026] Figure 4 This is a schematic diagram of the structure of the terminal of this application.

[0027] Component number description:

[0028] 1 DNA data module

[0029] 2 RNA Data Module

[0030] 3. First analysis module

[0031] 4 Second analysis module

[0032] 5. Judgment Module

[0033] 300 Terminal

[0034] 301 Processor

[0035] 302 Memory

[0036] 3021 Operating System

[0037] 3022 Applications

[0038] 304 Network Interface

[0039] 305 Bus system

[0040] 306 User Interface DETAILED DESCRIPTION

[0041] In order to make the invention purpose, technical scheme and beneficial effect of the present application clearer, the present application is further described below in conjunction with examples. It should be understood that the examples are only used to explain the present application and are not used to limit the scope of the application. The test methods used in the following examples are conventional methods unless otherwise specified, and are carried out according to the technology or conditions described in the literature in the art or according to the product specification. People familiar with this technology can easily understand other advantages and effects of the present application from the content disclosed in this description. The materials, reagents, etc. used in the following examples, unless otherwise specified, can all be obtained from commercial sources.

[0042] like Figure 1 As shown, the present application provides, on the one hand, a system for auxiliary diagnosis of benign and malignant thyroid nodules based on the combination of genomic variation and abnormal expression, comprising the following modules:

[0043] DNA data module 1, used to obtain data of sequencing libraries constructed by shingled multiplex amplification of DNA of each sample;

[0044] RNA data module 2, connected to the DNA data module 1, is used to obtain the data of the sequencing library constructed by multiple amplification after reverse transcription of the RNA of each sample and the RNA expression level data of the target genome of each sample: wherein the target genome includes: 14 characteristic genes and 7 internal reference genes; the 14 characteristic genes are AGGF1, BANP, CDK12, CITED1, DICER1, ERBB3, ERBB4, FGFR2, FGFR4, IGF2BP3, KIT, MET, VHL, ZBTB8A;

[0045] The first analysis module 3 is connected to the DNA data module 1 and the RNA data module 2, and is used to perform bioinformatics analysis on the data of the sequencing library obtained in the DNA data module 1 and the RNA data module 2, and obtain the benign or malignant judgment result of the thyroid nodule of each sample in combination with the somatic cell variation interpretation guide;

[0046] The second analysis module 4 is connected to the RNA data module 2, and is used to process and analyze the RNA expression level data of the target genome of each sample, and to predict the benign and malignant nature of the thyroid nodules for each sample based on the constructed benign and malignant nature prediction model to obtain the benign and malignant nature prediction result of the thyroid nodules of the corresponding sample;

[0047] The judgment module 5 is connected to the first analysis module 3 and the second analysis module 4, and is used to judge the final benign or malignant judgment result of each sample based on the benign or malignant judgment result of each sample and the benign or malignant prediction result of each sample.

[0048] In the system provided by the present application, the sample is selected from a fresh tissue sample of a thyroid nodule, a puncture biopsy cell sample, or a paraffin-embedded fixed tissue sample.

[0049] In the system provided by the present application, in the DNA data module 1, DNA comes from AKT1, ALK, APC, ATM, BANP, BRAF, CDK12, CDKN2A, CHEK2, CTNNB1, DICER1, EGFR, EIF1AX, EP300, ERBB4, EZH1, FAM193A, FARSB, FGFR1, FGFR2, FGFR3, FGFR4, FLT3, GLIS3, GNAQ, GNAS, HRAS, IDH1, IDH2, KIT, KL K1, KMT2C, KMT2D, KRAS, LRP1B, MEN1, MET, MTOR, NCOR2, NF1, NF2, NOTCH1, NRAS, NTRK1, NTRK3, PIK3CA, PLEKHS1, POR, PPARG, PTEN, PTH, RB1, RBM10, RET, ROS1, SMAD4, SPOP, STK11, SUGCT, TP53, TRIM61, TSC2, TSHR, VHL, ZNF148 and TERT genes. Among them, SUGCT is also called C7orf10.

[0050] In the system provided by the present application, in the RNA data module 2, the RNA comes from a fusion gene; the fusion gene is formed by the fusion of the upstream segment of the first gene breakpoint and the downstream segment of the second gene breakpoint;

[0051] There are 134 first genes in total. The first genes and their breakpoints are selected from the following: The numbers in brackets after the following genes indicate the exon numbers at the breakpoints:

[0052] AGGF1(5),AGK(2),AKAP13(35),AKAP9(21),AKAP9(7),AKAP9(8),AP3B1(22),ATIC(7),BCL2L11(3),CARS(17),CCDC30(11),CCDC6(1),CCDC6(2),CCD C6(8),CCNY(1),CD74(6),CLTC(30),CLTC(31),CREB3L2(2),DCTN1(16),DCTN1(26),EGFR(24),EML4(13),EML4(14),EML4(15),EML4(17),EML4(18), EML4(2),EML4(20),EML4(6),ERC1(11),ERC1(12),ETV6(4),ETV6(5),EZR(11),FAM114A2(9),FGFR1(1),FGFR1(14),FGFR1(17),FGFR1(2),FGFR2(17). ),FGFR3(17),FGFR3(18),FKBP15(9),FN1(23),GFPT1(19),GOLGA5(7),GOPC(4),GOPC(8),HIP1(21),HIP1(28),HIP1(30),HOOK3(11),IRF2BP2(1),K IAA1217(11),KIAA1549(12),KIAA1549(14),KIAA1549(15),KIAA1549(17),KIAA1549(18),KIAA1598(11),KIF5B(15),KIF5B(16),KIF5B(17),KIF5B (22),KIF5B(23),KIF5B(24),KLHL7(5),KTN1(29),LMNA(2),LRIG3(16),MACF1(60),MKRN1(4),MSN(11),NACC2(4),NCOA4(7),NCOA4(8),NPM1(4),PA X8(10),PAX8(2),PAX8(7),PAX8(8),PAX8(9),PCM1(29),POR(3),PPFIBP1(9),PRKAR1A(7),QKI(6),RANBP2(18),RBMS3(11),RBPMS(5),RELCH(10),R MDN3(7),RNF213(2),SDC4(2),SDC4(4),SEC31A(20),SHTN1(11),SLC34A2(13),SLC34A2(4),SND1(10),SND1(14),SND1(9),SPECC1L(10),SQSTM1(5),SSBP2(12),STRN(3),SYN2(5),TANK(4),TBL1XR1(9),TFG(4),TFG(5),TFG(6),THADA(28),THADA(29),THADA(30),THADA(31),THADA(35),THADA(36),TPM3(7) ,TPM3(8),TPR(15),TPR(21),TPR(6),TRIM24(10),TRIM24(3),TRIM24(5),TRIM24(8),TRIM24(9),TRIM27(3),TRIM33(16),UACA(17),VCL(16),ZC3HAV1(7);,

[0053] There are 48 second genes in total. The first gene and its breakpoint are selected from any of the following. The numbers in brackets after the following genes indicate the exon number at the breakpoint:

[0054] AFF3(7),ALK(19),ALK(20),BAIAP2L1(2),BICC1(10),BRAF(10),BRAF(11),BRAF(8),BRAF(9),CASP7(7),CCAR2(4),CCDC6(2),GLIS1( 2),GLIS3(3),IGF2BP3(2),IGF2BP3(3),LOC389473(1),LTK(10),MET(15),NTRK1(10),NTRK1(12),NTRK1(9),NTRK2(13),NTRK2(16),N TRK3(14),NTRK3(15),OFD1(3),PLAG1(2),PPARG(1),PPARG(2),RAD51(4),RAF1(9),RET(11),RET(12),ROS1(32),ROS1(34),ROS1(35) ,ROS1(36),SEPTIN14(10),SEPTIN14(8),SLC26A11(8),TACC1(7),TACC3(10),TACC3(11),TACC3(4),TACC3(8),TACC3(9),ZNF703(2).

[0055] In the system provided by the present application, in the DNA data module 1, the coverage of the shingled multiplex amplification is all exon regions of the DNA gene, and the region of at least 20 bases extending outward. When the DNA detection object is the TERT gene, due to the high correlation between its promoter region mutation and thyroid cancer, the coverage of the shingled multiplex amplification also includes its promoter region.

[0056] Since each DNA gene has multiple exons, only the detection regions of BRAF, RET and TP53 gene exons are listed here (using the reference genome coordinates of the GRCh37 version), see the table below:

[0057]

[0058]

[0059] Among them, Gene represents the gene name, RefSeq mRNA ID represents the reference transcript number (a unique identifier for a gene), Chrom represents the chromosome number, Exon_start-20bp represents the exon start coordinate extending outward by 20 bases, and Exon_end+20bp represents the exon end coordinate extending outward by 20 bases.

[0060] In the system provided by the present application, in the RNA data module 2, the scope of reverse transcription amplification is the expression transcript of the fusion gene.

[0061] In the system provided by the application, in DNA data module 1 or RNA data module 2, the primer used for amplification adopts NGS-PrimerPlex software design to obtain.The ratio of the primer used for amplification is determined by collecting primer characteristic data, based on vector regression model optimization.The ratio mode of the primer used for amplification is detailed in patent CN116092585A, which is characterized by collecting the features such as primer length and GC content, amplification region length and GC content, 5' end and 3' end Δ G, other regional similarities of the designed primers, and the features such as the amplicon quantity of other primers in the same group, the average GC content of the amplicon of other primer pairs, the average amplification region length of other primer pairs, the similarity of other primer pairs with the target primer pairs, based on certain test data resume support vector regression machine learning model, to determine the best primer concentration ratio.The ratio of primers is repeatedly adjusted based on multiple rounds of verification data, ensuring fabulous amplification uniformity.

[0062] In the system provided by the present application, in the DNA data module 1 or the RNA data module 2, when constructing the sequencing library, it includes connecting the target fragment to the index for sequencing. The index is selected from a single-end index and a double-end index; preferably, the index is a UDI; optionally, the index also includes a UMI (Unique Molecular Identifer). The sequencing connector is selected from the universal sequencing connector of Illumina or the sequencing connector of BGI. The sequencing platform is selected from the universal sequencing platform of Illumina or the sequencing platform of BGI. The sequencing method is selected from the reversible terminal termination sequencing method or the combined probe anchor polymerization sequencing method. Before sequencing, steps such as library purification and library quality control are also included to purify the library and ensure that the library concentration is qualified.

[0063] In the system provided by the present application, in RNA data module 2, the above 14 characteristic genes are selected because the abnormal changes in their expression levels can form an effective characteristic combination for predicting the benign and malignant nature of thyroid tissue.

[0064] In the system provided by the present application, in the first analysis module 3, the bioinformatics analysis can be implemented by a variety of software, including, for example, the software functions described below. The software names described below are all commonly used examples. There are many similar software that can achieve the same functions. The present invention does not specifically require the software with the following names:

[0065] The first analysis module 3 performs bioinformatics analysis on the data of the sequencing library obtained from the DNA data module 1 and the RNA data module 2 in the following manner:

[0066] Data quality control is performed on the data of the sequencing library corresponding to the DNA of each sample and the data of the sequencing library corresponding to the RNA; specifically, the raw sequencing data is quality controlled using Trimmomatic software, including checking the quality of the data, removing low-quality sequences, and removing PCR duplicate sequences. The sequencing reads that are generally filtered through this step are called cleanreads.

[0067] The data of the sequencing library corresponding to the DNA that has passed data quality control in each sample and the data of the sequencing library corresponding to the RA are compared with the DNA reference genome and the DNA reference genome to obtain the DNA gene comparison results and the RA gene comparison results; specifically, the high-quality sequencing data are compared with the reference genome using BWA software to determine the genomic position of each sequencing read length.

[0068] Based on the DNA gene comparison results and RNA gene comparison results of each sample, mutation detection and fusion detection are performed, and the mutation sites are filtered; GATK software is used to identify the mutations in the genome based on the comparison results. The types of mutations concerned by the present invention include single nucleotide variations (SNVs), short insertions and deletions (short indels), and gene fusions.

[0069] After BWA alignment, the fusion expression reference was successfully aligned. When the number of reads supporting a fusion was greater than or equal to 200 per 50M clean reads, it was judged to be positive for the fusion.

[0070] The main filtering objects in the present invention are the total sequencing depth (DP) of the detected sites, the number of supported reads (AD) that meet the variant alleles, and the variant frequency (VAF). The detected SNV / Indel is filtered by setting different DP, AD and VAF filtering standards for hot spots and non-hot spots, specifically DP>=50, AD>=30, VAF>=1% (hot spots) and DP>=100, AD>=50, VAF>=5% (non-hot spots). In addition, the site must meet the mutation frequency <0.001 (1000G_AF, 1000G_EAS_AF, gnomAD_AF, gnomAD_EAS_AF) in the healthy person database, and removes sites with chain preference (SOR>3) or sites that fall on the edge of the insert part of the amplicon where sequencing errors are prone to occur, and a blacklist set based on a large amount of R&D test data to filter out background mutations and possible sequencing errors.

[0071] Based on the annotation database, the filtered variant sites of each sample are annotated as detection sites, and their molecular biology and population information characteristics are marked; specifically, VEP software is used to annotate the filtered variants based on the annotation database, and their molecular biology and population information characteristics are marked.

[0072] In the system provided by the present application, in the first analysis module 3, the somatic variation interpretation guide is selected from the somatic variation interpretation guide formulated by the AMP / ASCO / CAP organization in the United States. The detected variant sites will be given one of the following five qualitative classifications: pathogenic (P, Pathogenic), likely pathogenic (LP, Likely Pathogenic), uncertain significance (VUS, variant of uncertain signifcance), likely benign (LB, Likely Benign), benign (B, Benign). This interpretation guide provides both the final judgment result of auxiliary diagnosis and an evaluation standard for the clinical effectiveness of the test method.

[0073] In the system provided in the present application, if the detection site of the sample corresponds to one or more of the three types of mutations including BRAFV600E, TERT promoter mutation and RET fusion gene, it is interpreted as pathogenic based on the somatic mutation interpretation guidelines.

[0074] In the system provided in the present application, the internal reference genes are selected from the expression profile data of cancer tissue and adjacent tissue of thyroid cancer patients in the TCGA database; the 7 internal reference genes are SLC5A5, B2M, PPIH, HPRT1, NONO, PUM1, and SDHA.

[0075] The specific selection method includes: selecting a group of internal reference genes from the expression profile data of cancer tissues and adjacent tissues of thyroid cancer patients in the TCGA database, whose expression levels meet the following conditions: (1) the coefficient of variation of expression in all samples is less than 20%; (2) the percentile of the average expression level in all gene transcripts is between 30% and 70%. The list of internal reference genes selected accordingly is: SLC5A5, B2M, PPIH, HPRT1, NONO, PUM1, SDHA.

[0076] In the system provided by this application, Figure 2 In the second analysis module 4, the method of constructing a prediction model for benign and malignant thyroid nodules includes:

[0077] The RNA expression level data of each characteristic gene and each internal reference gene in multiple benign nodule samples and multiple corresponding malignant papillary thyroid carcinoma samples without BRAF V600E, TERT promoter mutation and RET fusion were obtained; for example, 100 benign nodule samples and 100 malignant papillary thyroid carcinoma (PTC) samples without BRAF V600E, TERT promoter mutation and RET fusion were collected for modeling. This is because the malignant correlation of these three mutations is high enough and accounts for a high proportion in PTC, so the thyroid nodule benign and malignant prediction model is mainly used to evaluate the probability of benign and malignant without these three markers.

[0078] Processing and analyzing the RNA expression level data of the target genome of each sample to obtain processing and analysis results of multiple training samples;

[0079] The normal distribution of each characteristic gene was calculated based on the processing and analysis results of each training sample, and a prediction model for benign and malignant thyroid nodules was constructed in combination with the benign and malignant results of thyroid nodules of each training sample.

[0080] Specifically, the benign and malignant conclusions of each training sample are used as sample label values ​​(1abel). The data set obtained after the samples have been processed in the above process is randomly and equally divided into training set, validation set and test set to ensure that the distribution of malignant labels is representative in all sets. Use Python scikit-learn to train the random forest algorithm model. Set the model hyperparameters, including the number of trees, maximum tree depth and minimum samples per leaf. Use optimization methods to find the best hyperparameter values. Use the training data set to train the random forest model and monitor the performance of the model on the validation set to avoid overfitting. The model will output a floating point value in the range of (0, 1) as the predicted value of the model.

[0081] In the system provided in the present application, the processing and analysis of the RNA expression level data of the target genome of each sample includes:

[0082] Data quality control is performed on the RNA expression level data of each sample; specifically, the raw sequencing data is quality controlled using Trimmomatic software, including checking the quality of the data, removing low-quality sequences, and removing PCR duplicate sequences. Sequencing reads that are filtered through this step are generally called clean reads.

[0083] The RA expression level data of each sample that has passed data quality control are sequenced with the corresponding reference genome to obtain the number of reads of the corresponding gene amplicon of each gene in the target genome of each sample; based on the number of reads of the corresponding gene amplicon of each gene in the target genome of each sample, the RPKM value of each gene in the target genome of each sample is calculated; the ratio between the RPKM value of each characteristic gene in each sample and the total RPKM value of the internal reference gene is calculated to obtain the RPKM value ratio corresponding to each characteristic gene in each sample;

[0084] Based on the K-means clustering method, outlier samples are determined based on the RPKM value ratios corresponding to each characteristic gene in each sample, and outlier samples are excluded. Specifically, this step identifies samples with significantly large expression differences as outlier samples. Such samples may not be used for modeling and predictive analysis of RA expression due to problems such as sample quality, library construction, and sequencing quality, and such samples need to be excluded. The identification method of outlier samples is as follows:

[0085] Using the K-means clustering method, data points that are far from the centroid of their assigned cluster can be judged as outliers: the Euclidean distance of each point to the centroid of its cluster is calculated, and a threshold based on 2 standard deviations is set to judge outliers. The method includes the following steps: First, a predetermined number of clusters is selected using the Elbow method, denoted by "K". Each data point in the data set is assigned to its nearest centroid using the Euclidean distance. After the assignment is completed, the centroid of each cluster is recalculated as the average of all data points assigned to the cluster. The above assignment and centroid replacement process is repeated until the centroid position no longer changes. At this time, the distance of each data point to the centroid of its assigned cluster is calculated, and data points that exceed 2 times the standard error distance of the intra-cluster distance are judged as outliers.

[0086] In the system provided in the present application, the normal distribution of each characteristic gene is calculated based on the processing and analysis results of each training sample, including:

[0087] Based on the RPKM value ratio of each benign nodule sample in each training sample, the median RPKM value of each characteristic gene is obtained;

[0088] The log2 fold change value of each characteristic gene was calculated based on the RPKM value ratio of each characteristic gene in each benign nodule sample and the corresponding RPKM value median, and the normal distribution of each characteristic gene was obtained based on the median of the log2 fold change value of each characteristic gene and the absolute median difference.

[0089] In the system provided by the present application, the second analysis module 4 predicts the benign or malignant nature of thyroid nodules for each sample in the following manner:

[0090] The log2 fold change value of each characteristic gene was calculated by comparing the RPKM value ratio corresponding to each characteristic gene of each sample with the corresponding RPKM value median;

[0091] Based on the normal distribution of each characteristic gene, the gene expression variation results of each characteristic gene of each sample were obtained according to the calculated log2 fold change value of each characteristic gene;

[0092] Based on the gene expression results of each characteristic gene of each sample, the prediction results of benign or malignant thyroid nodules corresponding to each sample are obtained.

[0093] In the system provided in the present application, the method of obtaining the gene expression variation result of each characteristic gene of each sample according to the calculated log2 fold change value of each characteristic gene includes:

[0094] Based on the gene expression variation result judgment rule, determine whether the calculated log2 fold change value of each characteristic gene is within the 95% confidence interval of the normal distribution of the corresponding characteristic gene, and obtain the gene expression variation result of each characteristic gene of each sample;

[0095] If it is within the 95% confidence interval of the normal distribution, it is judged that the corresponding characteristic gene has not mutated;

[0096] If it is outside the left side of the 95% confidence interval of the normal distribution, the corresponding characteristic gene is judged to be a low-expression variation;

[0097] If it is outside the right side of the 95% confidence interval of the normal distribution, the corresponding characteristic gene is judged to be overexpressed.

[0098] In the system provided by the present application, the judgment module 5 judges the final benign or malignant judgment result of each sample based on the benign or malignant judgment result of each sample and the benign or malignant prediction result of each sample, including:

[0099] If the benign or malignant judgment result of the thyroid nodule of the sample corresponds to pathogenic or possibly pathogenic, the final benign or malignant judgment result of the corresponding sample is judged to be malignant;

[0100] If the benign or malignant judgment result of the thyroid nodule of the sample is unclear, possibly benign, or benign, the final benign or malignant judgment result of the thyroid nodule of the corresponding sample obtained based on the benign or malignant prediction result of the thyroid nodule of each sample is judged to be malignant.

[0101] In a specific embodiment, the method of judging the final benign or malignant thyroid nodule judgment result of each sample based on the benign or malignant thyroid nodule judgment result of each sample and the benign or malignant thyroid nodule prediction result includes:

[0102] a. For samples containing three types of mutations, BRAFV600E, TERT promoter mutation, and RET fusion gene, in order to clarify the highly malignant mutations, the interpretation of the benign and malignant judgment results of thyroid nodules is definitely pathogenic (P). There is no need to further interpret the benign and malignant prediction results of thyroid nodules.

[0103] b. For samples containing other types of mutations, if at least one of the following two conditions is met, it is judged to be malignant:

[0104] i. Based on the AMP / ASCO / CAP somatic variant interpretation guidelines, the benign or malignant nature of thyroid nodules is determined as pathogenic (P) or possibly pathogenic (LP);

[0105] ii. Based on the final benign or malignant judgment of the thyroid nodule, it is judged to be malignant.

[0106] It should be noted that the system for auxiliary diagnosis of benign and malignant thyroid nodules based on the combination of genomic variation and abnormal expression provided in the above embodiment only uses the division of the above-mentioned program modules as an example when performing auxiliary diagnosis of benign and malignant thyroid nodules. In actual applications, the above-mentioned processing can be assigned to different program modules as needed, that is, the internal structure of the system can be divided into different program modules to complete all or part of the processing described above.

[0107] On the other hand, the present application provides a method for auxiliary diagnosis of benign and malignant thyroid nodules for non-disease diagnosis purposes, comprising the following steps:

[0108] 1) Obtaining data of sequencing libraries constructed by shingled multiplex amplification of DNA from each sample;

[0109] 2) Obtaining RNA from each sample for reverse transcription and multiple amplification to construct sequencing library data and RA expression level data of the target genome of each sample; wherein the target genome includes: 14 characteristic genes and 7 internal reference genes; the 14 characteristic genes are AGGF1, BANP, CDK12, CITED1, DICER1, ERBB3, ERBB4, FGFR2, FGFR4, IGF2BP3, KIT, MET, VHL, and ZBTB8A;

[0110] 3) Performing bioinformatics analysis on the sequencing library data of step 1) and step 2), and combining the somatic cell variation interpretation guide to obtain the benign or malignant judgment result of the thyroid nodule;

[0111] 4) Processing and analyzing the RNA expression level data of the target genome in step 2), and obtaining a prediction result of benign and malignant thyroid nodules based on the constructed prediction model of benign and malignant thyroid nodules;

[0112] 5) Based on the benign or malignant judgment result of the thyroid nodule in step 3) and the benign or malignant prediction result of the thyroid nodule in step 4), a final benign or malignant judgment result of the thyroid nodule is obtained.

[0113] In the method provided in the present application, the non-disease diagnosis purpose may be, for example, statistical analysis of genome sequence characteristics or variation information.

[0114] In the method provided in the present application, step 1) refers to obtaining the data of the sequencing library constructed by the shingled multiplex amplification of the DNA of each sample. The DNA is the DNA in the DNA data module in the aforementioned system. The method of constructing the sequencing library by shingled multiplex amplification is the same as the method of constructing the sequencing library in the DNA data module 1 in the aforementioned system.

[0115] In the method provided by the present application, step 2) refers to obtaining the RNA of each sample for reverse transcription and then multiple amplification to construct sequencing library data and RA expression level data of the target genome of each sample; wherein the target genome includes: 14 characteristic genes and 7 internal reference genes; the 14 characteristic genes are AGGF1, BANP, CDK12, CITED1, DICER1, ERBB3, ERBB4, FGFR2, FGFR4, IGF2BP3, KT, MET, VHL, ZBTB8A. wherein RNA is the RNA in the RNA data module in the aforementioned system. the range of reverse transcription amplification of RA is the same as the range of reverse transcription amplification in RA data module 2 in the aforementioned system.

[0116] In the method provided in the present application, in step 1) or step 2), the sample is selected from a fresh tissue sample of a thyroid nodule, a puncture biopsy cell sample, or a paraffin-embedded fixed tissue sample. In step 1) or step 2), the primer design, primer ratio, and sequencing library construction method used for amplification correspond to the DNA data module 1 or RNA data module 2 in the aforementioned system, respectively, and will not be repeated here.

[0117] In the method provided in the present application, step 3) refers to performing bioinformatics analysis on the data of the sequencing library of step 1) and step 2), and combining the somatic cell variation interpretation guide to obtain the benign and malignant judgment result of the thyroid nodule. Wherein, the somatic cell variation interpretation guide is selected from the somatic cell variation interpretation guide formulated by the AMP / ASCO / CAP organization in the United States; preferably, if the detection site of the sample corresponds to one or more of the three types of variation including BRAF V600E, TERT promoter variation and RET fusion gene, it is interpreted as pathogenic based on the somatic cell variation interpretation guide.

[0118] In the method provided in the present application, step 4) refers to processing and analyzing the RNA expression level data of the target genome in step 2), and obtaining the prediction result of benign and malignant thyroid nodules based on the constructed benign and malignant thyroid nodule prediction model. Among them, the process of processing and analyzing the RNA expression level data and the method of constructing the benign and malignant thyroid nodule prediction model are consistent with the process in the second analysis module 4 in the aforementioned system, and will not be repeated here.

[0119] In the method provided in the present application, step 5) refers to obtaining the final benign and malignant judgment result of the thyroid nodule based on the benign and malignant judgment result of the thyroid nodule in step 3) and the benign and malignant prediction result of the thyroid nodule in step 4). The method of judging the final benign and malignant judgment result of each sample based on the benign and malignant judgment result of the thyroid nodule and the benign and malignant prediction result of the thyroid nodule includes:

[0120] If the benign or malignant judgment result of the thyroid nodule of the sample corresponds to pathogenic or possibly pathogenic, the final benign or malignant judgment result of the corresponding sample is judged to be malignant;

[0121] If the benign or malignant judgment result of the thyroid nodule of the sample is unclear, possibly benign, or benign, the final benign or malignant judgment result of the thyroid nodule is obtained based on the benign or malignant prediction result of the thyroid nodule of each sample.

[0122] On the other hand, the present application provides a computer-readable storage medium having a computer program stored thereon, which implements the analysis steps of step 3) to step 5) in the aforementioned method when the computer program is executed by a processor.

[0123] Another aspect of the present application provides a terminal, including a processor and a memory;

[0124] The memory is used to store computer programs;

[0125] The processor is used to execute the computer program stored in the memory, so that the device performs the analysis steps of step 3) to step 5) in the aforementioned method.

[0126] The method for auxiliary diagnosis of benign and malignant thyroid nodules provided in the embodiment of the present application for non-disease diagnosis purposes can be implemented on the terminal side or the server side. As for the hardware structure of the terminal, please refer to Figure 4 , is an optional hardware structure diagram of a terminal 300 provided in an embodiment of the present invention. The terminal 300 may be a mobile phone, a computer device, a tablet device, a personal digital processing device, a factory background processing device, etc. The terminal 300 includes: at least one processor 301, a memory 302, at least one network interface 304 and a user interface 306. The various components in the device are coupled together through a bus system 305. It can be understood that the bus system 305 is used to realize the connection and communication between these components. In addition to the data bus, the bus system 305 also includes a power bus, a control bus and a status signal bus. However, for the sake of clarity, in Figure 4 In the specification, various buses are labeled as bus systems.

[0127] The user interface 306 may include a display, a keyboard, a mouse, a trackball, a click gun, keys, buttons, a touch pad or a touch screen.

[0128] It is understood that the memory 302 can be a volatile memory or a non-volatile memory, and can also include both volatile and non-volatile memories. Among them, the non-volatile memory can be a read-only memory (ROM), a programmable read-only memory (PROM), which is used as an external cache. By way of example but not limitation, many forms of RAM are available, such as static random access memory (SRAM), synchronous static random access memory (SSRAM). The memory described in the embodiments of the present invention is intended to include but is not limited to these and any other suitable categories of memory.

[0129] The memory 302 in the embodiment of the present invention is used to store various categories of data to support the operation of the terminal 300. Examples of these data include: any executable program operating on the terminal 300, such as an operating system 3021 and an application 3022; the operating system 3021 includes various system programs, such as a framework layer, a core library layer, a driver layer, etc., for implementing various basic services and processing hardware-based tasks. The application 3022 may include various applications, such as a media player (MediaPlayer), a browser (Browser), etc., for implementing various application services. The method for auxiliary diagnosis of benign and malignant thyroid nodules provided in the embodiment of the present invention may be included in the application 3022.

[0130] The method disclosed in the above embodiment of the present invention can be applied to the processor 301, or implemented by the processor 301. The processor 301 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the hardware integrated logic circuit in the processor 301 or the instruction in the form of software. The above processor 301 may be a general processor, a digital signal processor (DSP, Digital Signal Processor), or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. The processor 301 can implement or execute the methods, steps and logic block diagrams disclosed in the embodiment of the present invention. The general processor 301 can be a microprocessor or any conventional processor, etc. In combination with the steps of the accessory optimization method provided in the embodiment of the present invention, it can be directly embodied as a hardware decoding processor to execute, or it can be executed by a combination of hardware and software modules in the decoding processor. The software module can be located in a storage medium, which is located in a memory, and the processor reads the information in the memory and completes the steps of the above method in combination with its hardware.

[0131] In an exemplary embodiment, the terminal 300 may be implemented by one or more application specific integrated circuits (ASIC), DSP, programmable logic device (PLD), complex programmable logic device (CPLD) to execute the aforementioned method.

[0132] Those skilled in the art can understand that all or part of the steps of implementing the above-mentioned method embodiments can be completed by hardware related to the computer program. The aforementioned computer program can be stored in a computer-readable storage medium. When the program is executed, the steps of the above-mentioned method embodiments are executed; and the aforementioned storage medium includes: ROM, RAM, magnetic disk or optical disk and other media that can store program codes.

[0133] In the embodiments provided in the present application, the computer readable and writable storage medium may include a read-only memory, a random access memory, an EEPROM, a CD-ROM or other optical disk storage device, a disk storage device or other magnetic storage device, a flash memory, a USB flash drive, a mobile hard disk, or any other medium that can be used to store a desired program code in the form of an instruction or data structure and can be accessed by a computer. In addition, any connection can be appropriately referred to as a computer-readable medium. For example, if the instruction is sent from a website, a server or other remote source using a coaxial cable, an optical fiber cable, a twisted pair, a digital subscriber line (DSL), or wireless technologies such as infrared, radio, and microwaves, the coaxial cable, optical fiber cable, twisted pair, DSL, or wireless technologies such as infrared, radio, and microwaves are included in the definition of the medium. However, it should be understood that computer readable and writable storage media and data storage media do not include connections, carriers, signals, or other temporary media, but are intended to be non-temporary, tangible storage media. Disk and disc, as used in this application, includes compact disc (CD), laser disc, optical disc, digital versatile disc (DVD), floppy disk and Blu-ray disc where disks usually reproduce data magnetically, while discs reproduce data optically with lasers.

[0134] Based on the detection of genomic variation, the present invention combines the detection of nodule tissue transcriptional expression markers and provides a set of quantitative, highly scalable and quantifiable benign and malignant analysis methods, which has important application value in clinical practice.

[0135] The present application is further described below by way of examples, but the scope of the present application is not limited thereby. The most suitable detection object of the present invention is a thyroid nodule puncture tissue sample, and it can also be used for other purposes such as. The use process is basically sample processing, NGS library construction, sequencing and analysis, and data interpretation. The following is a detailed explanation of this process:

[0136] Example 1

[0137] 1. Nucleic Acid Extraction

[0138] Fresh tissue samples, fine needle aspiration samples, or formalin-fixed paraffin-embedded (FFPE) samples of thyroid nodules were collected from the clinic and analyzed using QIAGEN DNA / RNA Kit or DNA / RNA FFPE Kit, extract DNA and RNA from the sample according to the instructions. Then use Nanodrop spectrophotometer and Qubit fluorometer to detect the purity and concentration of nucleic acid.

[0139] 2. RNA Reverse Transcription

[0140] This step reverse transcribes the sample RNA into first-strand cDNA.

[0141] 2.1 Prepare the reverse transcription reaction solution on ice:

[0142] Components volume Template RNA (5~80ng) xμL RT Master Mix 1.6μL Nuclease-free Water to 8μL Total 8μL

[0143] Note: (1) The recommended input amount of Template RNA for non-FFPE samples is 40 ng, and the recommended input amount for FFPE samples is 80 ng.

[0144] (2) Template RNA input volume x (μL) = sample input amount (ng) / sample concentration (ng / μL).

[0145] 2.2 Use a pipette to gently pipette and centrifuge briefly. Place the mixture in a PCR instrument and perform the first-strand cDNA synthesis reaction under the following conditions (heated cover at 105°C):

[0146] temperature time 25℃ 10min 42℃ 40min 85℃ 5min 12℃ Hold

[0147] Note: The synthesized cDNA product can be directly used in the next PCR reaction.

[0148] 3. Multiplex PCR Amplification of Target Genes

[0149] This step performs specific PCR amplification of the target gene.

[0150] 3.1 Prepare multiplex PCR reaction solution on ice:

[0151] In the DNA Panel, taking the BRAF gene as an example, the table below lists a design method for all primer pairs covering all target detection regions within the BRAF gene.

[0152]

[0153]

[0154] In RNA Panel, taking the common 3' driver gene RET as an example, the following table lists all RET as the 3' gene fusion detection target combinations. The table shows the 5' gene name, the exon number at the 5' breakpoint, the 3' gene name, the exon number at the 3' breakpoint, the 5' amplification primer sequence, and the 3' amplification primer sequence.

[0155]

[0156]

[0157] The reaction system is as follows:

[0158]

[0159]

[0160] Components (DNA Pool 1 reaction system) volume Template DNA (2.5~20ng) xμL 1st PCR Mix 5μL DNA Primer-1 1.3μL Nuclease-free Water to 10μL Total 10μL

[0161] Components (DNA Pool 2 reaction system) volume Template DNA (2.5~20ng) xμL 1st PCR Mix 5μL DNA Primer-2 1.3μL Nuclease-free Water to 10μL Total 10μL

[0162] Note: (1) For each DNA sample, two reaction systems, Pool 1 and Pool 2, must be prepared simultaneously, with DNA Primer-1 and DNA Primer-2 added to each system.

[0163] (2) It is recommended that the amount of Template DNA input into each pool is 10 ng for non-FFPE samples and 20 ng for FFPE samples.

[0164] (3) Template DNA input volume x (μL) = sample input amount (ng) / sample concentration (ng / μL).

[0165] 3.2 Use a pipette to gently pipette and mix, centrifuge briefly, and run the following program on a PCR instrument (heat cover 105°C):

[0166]

[0167] 4. PCR Product Purification

[0168] 4.1Beads were equilibrated at room temperature for 30 min and vortexed to mix.

[0169] 4.2 Add 10 μL Low TE Buffer to each of the DNA Pool 1 and DNA Pool 2 reaction tubes to make up to 20 μL.

[0170] 4.3 Pipette 16 μL (0.8×) beads into DNA Pool 1, DNA Pool 2 and RNA reaction tubes, mix by pipetting or vortexing gently, and incubate at room temperature for 5 min.

[0171] 4.4 Briefly centrifuge the PCR tube and place it on a magnetic stand to separate the magnetic beads and liquid. After the solution is clarified, transfer the supernatant to a PCR tube containing 10 μL (0.5×) Beads, mix by pipetting or gently vortexing, and incubate at room temperature for 5 minutes.

[0172] 4.5 Briefly centrifuge the PCR tube and place it on a magnetic rack to separate the magnetic beads and liquid. After the solution is clear, discard the supernatant.

[0173] 4.6 Keep the PCR tube on the magnetic rack, add 200 μL of freshly prepared 70% ethanol to rinse the magnetic beads, allow the magnetic beads to migrate in opposite directions 10 times, and discard the supernatant. Repeat this step once, rinsing twice in total.

[0174] 4.7 Briefly centrifuge the PCR tube and place it on a magnetic rack to discard the residual ethanol. Open the lid and dry until no ethanol remains (the magnetic beads do not reflect light).

[0175] Note: The magnetic beads should not be over-dried to the point of cracking.

[0176] 4.8 Add 15 μL Low TE Buffer to the RNA reaction tube, mix by pipetting or vortexing lightly, and centrifuge briefly. Add 7.5 μL Low TE Buffer to the DNA Pool 1 and DNA Pool 2 reaction tubes respectively, mix by pipetting or vortexing lightly, and centrifuge briefly to mix Pool 1 and Pool 2 corresponding to each DNA sample into one tube.

[0177] The purified PCR product (containing magnetic beads) was obtained.

[0178] 5. Enzymatic hydrolysis of non-specific products

[0179] This step is to remove non-target fragments generated by non-specific PCR amplification by enzymatic digestion.

[0180] 5.1 Prepare enzymatic reaction solution on ice:

[0181] Components volume Purified PCR products 15μL Digestion Buffer 2μL Digestion Enzyme 2μL Nuclease-free Water 1μL Total 20μL

[0182] 5.2 Use a pipette to gently mix, centrifuge briefly, and perform enzymatic reaction under the following conditions (heat cover 105℃):

[0183] temperature time 37℃ 25min 12℃ Hold

[0184] 6. Purification of enzymatic hydrolysate

[0185] 6.1Beads were equilibrated at room temperature for 30 min and vortexed to mix.

[0186] 6.2 Pipette 26 μL (1.3×) Beads into the enzymatic hydrolyzate, mix by pipetting or gently vortexing, and incubate at room temperature for 5 min.

[0187] 6.3 Briefly centrifuge the PCR tube and place it on a magnetic rack to separate the magnetic beads and liquid. After the solution is clear, discard the supernatant.

[0188] 6.4 Keep the PCR tube on the magnetic stand, add 200 μL of freshly prepared 70% ethanol to rinse the magnetic beads, allow the magnetic beads to migrate in opposite directions 10 times, and discard the supernatant. Repeat this step once, rinsing twice in total.

[0189] 6.5 Briefly centrifuge the PCR tube and place it on a magnetic rack to discard the residual ethanol. Open the lid and dry until no ethanol remains (the magnetic beads do not reflect light).

[0190] Note: The magnetic beads should not be over-dried to the point of cracking.

[0191] 6.6 Add 10 μL Low TE Buffer, mix by pipetting or vortexing, and centrifuge briefly to obtain the purified enzymatic product (containing magnetic beads).

[0192] 7. Library PCR Reaction

[0193] In this step, the target fragment is amplified by PCR and connected to a double-end index.

[0194] 7.1 Prepare library PCR reaction solution on ice:

[0195] Components volume Purified enzymatic hydrolysate 10μL 2nd PCR Mix 20μL UDI-XXX 4μL Nuclease-free Water 6μL Total 40μL

[0196] 7.2 Use a pipette to gently pipette and mix, centrifuge briefly, and run the following program on a PCR instrument (heat cover at 105°C):

[0197]

[0198] 8. Library Purification

[0199] 8.1Beads were equilibrated at room temperature for 30 min and vortexed to mix.

[0200] 8.2 Pipette 40 μL (1×) Beads into the PCR product, mix by pipetting or gently vortexing, and incubate at room temperature for 5 minutes.

[0201] 8.3 Briefly centrifuge the PCR tube and place it on a magnetic rack to separate the magnetic beads and liquid. After the solution is clear, discard the supernatant.

[0202] 8.4 Keep the PCR tube on the magnetic rack, add 200 μL of freshly prepared 70% ethanol to rinse the magnetic beads, allow the magnetic beads to migrate in opposite directions 10 times, and discard the supernatant. Repeat this step once, rinsing twice in total.

[0203] 8.5 Briefly centrifuge the PCR tube and place it on a magnetic rack to discard the residual ethanol. Open the lid and dry until no ethanol remains (the magnetic beads do not reflect light).

[0204] Note: The magnetic beads should not be over-dried to the point of cracking.

[0205] 8.6 Add 20 μL Low TE Buffer, mix by pipetting or gently vortexing, and incubate at room temperature for 5 min.

[0206] 8.7 After a brief centrifugation, place the PCR tube on a magnetic rack and pipette the supernatant into a new centrifuge tube to obtain a purified library. The library can be directly sequenced after quality inspection or stored at -20°C.

[0207] 9. Library quality control

[0208] 9.1 Use Qubit Fluorometer to measure the library concentration. The library concentration is qualified when it is ≥2ng / μL.

[0209] 9.2 Use agarose gel electrophoresis or Qsep100 fully automatic nucleic acid protein analysis system or Agilent2100Bioanalyzer system (High Sensitivity DNA Kit) to perform library fragment quality control. The length of the library fragment is distributed between 220 and 280 bp, and there is no obvious primer dimer (length between 160 and 190 bp).

[0210] 10. Sequencing

[0211] The library was sequenced using an Illumina gene sequencer according to the instrument's standard operating procedures. The recommended sequencing data volume is 2 Gb for DNA libraries and 1 Gb for RNA libraries.

[0212] 11. Bioinformatics Analysis

[0213] a. Data quality control: Use Trimmomatic software to perform quality control on raw sequencing data, including checking data quality, removing low-quality sequences, and removing PCR duplicate sequences. Sequencing reads that are filtered through this step are generally called clean reads.

[0214] b. Sequence alignment: Use BWA software to align high-quality sequencing data with the reference genome to determine the genomic location of each sequencing read.

[0215] c. Mutation detection: Using GATK software, based on the alignment results, the mutations in the genome are identified. The types of mutations that the present invention is concerned with include single nucleotide variants (SNVs), short insertions and deletions (short indels), and gene fusions.

[0216] d. Fusion detection: After BWA alignment, the fusion expression reference is successfully aligned. When the number of reads supporting a fusion is greater than or equal to 200 per 50M clean reads, it is judged as positive for the fusion.

[0217] e. The main filtering objects in the present invention are the total sequencing depth (DP) of the detected sites, the number of supported reads (AD) that meet the variant alleles, and the variant frequency (VAF). The detected SNV / Indel is filtered according to different DP, AD and VAF filtering standards set for hot spots and non-hot spots, specifically DP>=50, AD>=30, VAF>=1% (hot spots) and DP>=100, AD>=50, VAF>=5% (non-hot spots). In addition, the site must meet the mutation frequency <0.001 (1000G_AF, 1000G_EAS_AF, gnomAD_AF, gnomAD_EAS_AF) in the healthy person database, and the sites with chain preference (SOR>3) or the marginal positions where sequencing errors are prone to occur in the insert part of the amplicon are removed, and the blacklist set based on a large amount of R&D test data is used to filter out background mutations and possible sequencing errors.

[0218] f. Annotation and analysis: Use VEP software to annotate the filtered variants based on the annotation database and mark their molecular biology and population information characteristics.

[0219] g. RNA expression data were analyzed according to the following process:

[0220] h. Perform data quality control on RNA expression level data;

[0221] i. Use BWA software to compare the high-quality RNA expression level data that have undergone data quality control with the reference genome to obtain the number of reads of the corresponding gene amplicon;

[0222] j. The number of reads mapped to the corresponding gene amplicons was used as the gene expression to calculate the RPKM values ​​of 14 characteristic genes and 7 internal reference genes; the 14 characteristic genes were AGGF1, BANP, CDK12, CITED1, DICER1, ERBB3, ERBB4, FGFR2, FGFR4, IGF2BP3, KIT, MET, VHL, and ZBTB8A; the 7 internal reference genes were SLC5A5, B2M, PPIH, HPRT1, NONO, PUM1, and SDHA.

[0223] k. Expression normalization: Use the total RPKM value of these characteristic genes in each sample to normalize the RPKM values ​​of all other detected genes.

[0224] 1. After that, samples with significantly large expression differences need to be identified as outlier samples. Such samples may not be used for RNA expression modeling and predictive analysis due to problems such as sample quality, library construction, and sequencing quality. Such samples need to be excluded. The identification method of outlier samples is as follows:

[0225] m. Using the K-means clustering method, data points that are far away from the centroid of their assigned cluster can be judged as outliers: the Euclidean distance of each point to the centroid of its cluster is calculated, and a threshold based on 2 times the standard deviation is set to judge outliers. The method includes the following steps: First, a predetermined number of clusters is selected using the Elbow method, represented by "K". Each data point in the data set is assigned to its nearest centroid using the Euclidean distance. After the assignment is completed, the centroid of each cluster will be recalculated as the average of all data points assigned to the cluster. Repeat the above assignment and centroid replacement process until the centroid position no longer changes. At this time, the distance from each data point to the centroid of its assigned cluster is calculated, and data points that exceed 2 times the standard error distance of the intra-cluster distance are judged as outliers.

[0226] n. The above data can be used for modeling and model-based predictive analysis. First, a classification model (hereinafter referred to as Gene Expression Alteration Classifer, or GEA Classifer) is established using the random forest method and applied to the prediction of benign and malignant thyroid cancer tissue samples. The model is established as follows:

[0227] a. Select the sample set that meets the threshold of characteristic genes and removes outlier samples for establishing the expression model. A total of 100 benign nodule samples and 100 malignant papillary thyroid carcinoma (PTC) samples without BRAF V600E, TERT promoter mutation, and RET fusion were collected for modeling. This is because the malignant correlation of these three mutations is high enough and accounts for a high proportion in PTC, so this RNA expression module is mainly used to evaluate the probability of benign and malignant without these three markers.

[0228] b. For each characteristic gene in the expression model, the median of the RPKM value of the expression amount of adjacent cancer and benign samples is used as the baseline, and the log2 foldchange value of the characteristic gene in each sample relative to the baseline is calculated, and the median and the median absolute difference (MAD) of the log2fold change of adjacent cancer samples are used to simulate the normal distribution of each characteristic gene. The characteristics of this expression model are the classification of the expression values ​​of the samples in each gene, and the classification includes: -1, 0 and 1. -1 means that the log2 fold change value is outside the 95% confidence interval of the distribution to the left, 0 means that the log2 fold change value is within the 95% confidence interval of the distribution, and 1 means that the log2 fold change value is outside the 95% confidence interval of the distribution to the right.

[0229] c. Use the benign and malignant conclusions of the postoperative cases of the sample patients as sample labels (1abel).

[0230] d. The data set obtained after the samples have been processed in the above process is randomly and equally divided into a training set, a validation set, and a test set to ensure that the distribution of malignant labels is representative in all sets.

[0231] e. Use Python scikit-learn to train the random forest algorithm model.

[0232] f. Set model hyperparameters, including the number of trees, maximum tree depth, and minimum samples per leaf. Use optimization methods to find the best hyperparameter values.

[0233] g. Train the random forest model using the training dataset and monitor the performance of the model on the validation set to avoid overfitting.

[0234] h. The model will output a floating point value in the range of (0, 1), which is the predicted value of the model.

[0235] i. Use the test dataset to evaluate the performance of the model and calculate relevant indicators including accuracy, precision, recall, and F1 score. Figure 3As shown in the figure, the AUC of the model can reach 0.87, indicating that the prediction accuracy of the model for malignant samples without BRAF V600E, TERT promoter mutation, and RET fusion gene is as high as 0.87. The position with the largest Youden index is selected as the positive judgment value, and the output model prediction value is greater than the positive judgment value, which is judged as "malignant".

[0236] 12. Interpret the results based on the American AMP / ASCO / CAP interpretation guidelines. Specific methods:

[0237] a. Experienced readers collect the following relevant information based on the detected mutation results: the mutation characteristics of the site (such as whether it is a mistranslation mutation, a termination mutation, an insertion or deletion, an indirect mutation affecting the end of the exon, etc.), the frequency of the mutation in the population database, the mutation of the site in the malignant data points in the hotspot database, the summary results of existing functional studies, and the calculation method to predict its contribution to malignancy.

[0238] Specifically:

[0239] 1) The population database values ​​are derived from the frequency of the variant in the population based on the gNOMAD and 1000Genomes databases, which have clear quantitative standards and can be predefined and automated;

[0240] 2) Functional databases do not have definite indicators, which come from evidence based on conclusions in specific literature and cannot be predefined or automated;

[0241] 3) Prediction data predicts the possible degree of harm based on the molecular biological characteristics of the variant, which is a qualitative result. For example, if the missense mutation at the same position is a known pathogenic mutation, it is qualitatively classified as OM3, and the functional inactivation mutation of the tumor suppressor gene is defined as OVS1, etc. This type of interpretation often has highly clear standards and can basically be predefined and automated;

[0242] 4) Hotspot database: Data from the cancerhotspots.org database, based on the number of related records, can be predefined and automated;

[0243] 5) Additional supporting evidence provided by computer software: Specific software can be automated to support the prediction of pathogenicity.

[0244] b. For each of the above information points, the SOP provides a level of evidence strength and a corresponding score. It is divided into very strong (8 points), strong (4 points), moderate (2 points), and supporting (1 point). If the evidence supports malignant localization, it will be scored positively, and if it supports benign localization, it will be scored negatively.

[0245] c. Add up all the evidence scores designed for each variant site as the comprehensive interpretation score of the variant site.

[0246] d. Qualitative grading: 10 points or more are pathogenic (P), 6-9 points are possibly pathogenic (LP), 0-5 points are variant of uncertain significance (VUS), -1 to -6 points are probably benign (LB), and -7 points or less are benign (B).

[0247] e. Prediction data predicts the possible degree of harm based on the molecular biological characteristics of the variant, which is a qualitative result. For example, if the missense mutation at the same position is a known pathogenic mutation, it is qualitatively classified as OM3, and the functional inactivation mutation of the tumor suppressor gene is defined as OVS1, etc. This type of interpretation often has highly clear standards and can basically be predefined and automated;

[0248] f. Hotspot database: Data from the cancerhotspots.org database, based on the number of related records, can be predefined and automated;

[0249] g. Additional supporting evidence provided by computer software: Predictions supporting pathogenicity can be automated using specific software.

[0250] h. For each of the above information points, the SOP provides a level of evidence strength and a corresponding score. It is divided into very strong (8 points), strong (4 points), moderate (2 points), and supporting (1 point). If the evidence supports malignant localization, it will be scored positively, and if it supports benign localization, it will be scored negatively.

[0251] i. Add up all the evidence scores designed for each variant site as the comprehensive interpretation score of the variant site.

[0252] j. Qualitative grading: 10 points or more are pathogenic (P), 6-9 points are likely pathogenic (LB), 0-5 points are variant of uncertain significance (VUS), -1 to -6 points are likely benign (LB), and -7 points or less are benign (B).

[0253] For each detected variation, the qualitative criteria for benign or malignant are calculated by checking whether the variation meets the following characteristics:

[0254] 1) The strength of the evidence for pathogenicity is “very strong” (8 points), including:

[0255] 1.1. Functional inactivation variants in a confirmed tumor suppressor gene (nonsense mutation, frameshift mutation, classical ±1 or 2 splice site, start codon, single or multiple exon deletion). Code name OVS1.

[0256] 2) The strength of the evidence for pathogenicity is “strong” (4 points), including:

[0257] 2.1. Has the same amino acid change as a pathogenic variant previously determined by an expert, such as a different nucleotide change in the same codon. Code name OS1.

[0258] 2.2. The variant has been well-established to have an oncogenic effect in in vitro or in vivo functional studies. Code OS2; if OS1 has been adopted, this rule can only be used if the functional studies are based on specific nucleotide changes in the variant.

[0259] 2.3. Located in the cancerhotspots.org hotspot mutation database, at least 50 samples are recorded with somatic variants at the same amino acid position, and at least 10 samples in cancerhotspots.org have the same amino acid changes. If the tumor type where the somatic variant is located is not fully covered in cancerhotspots.org, resources such as COSMIC or tumor type-specific studies can be used. Code OS3; if OS1 has been adopted, this rule can only be used when the hotspot is clearly based on the specific nucleotide changes of the variant.

[0260] 3) The strength of the evidence for pathogenicity is “moderate” (2 points), including:

[0261] 3.1. The variant is located in a critical and well-defined functional domain (e.g., the active site of an enzyme). Code OM1; if OS1 or OS3 has been adopted, this rule cannot be used.

[0262] 3.2. Protein length changes caused by whole-frame deletions / insertions in known oncogenes or tumor suppressor genes, or by stop-loss variants in known tumor suppressor genes. Codename OM2.

[0263] 3.3. A missense variant, and a different missense variant that is confirmed to be carcinogenic has been recorded at the same amino acid. Code OM3; if OS1 or OS3 or OM1 has been adopted, this rule cannot be used.

[0264] 3.4. Located in one of the hotspots in cancerhotspots.org, with somatic variants in at least 10 samples at the same amino acid position, and the same amino acid change count in cancerhotspots.org is less than 50. If the tumor type where the somatic variant is located is not sufficiently covered by cancerhotspots.org, resources such as COSMIC or tumor type-specific studies can be used. Code OM4; if OM1 or OM3 has been adopted, this rule cannot be used.

[0265] 4) The strength of the evidence for pathogenicity is “supporting” (1 point), which includes:

[0266] 4.1. The computational algorithm obtains multiple pieces of predictive evidence supporting the oncogenic effect of the variant, including conservation / evolution, effects on splicing, etc. No matter how many algorithms are used for evaluation, the rule can only be calculated once. Code OP1.

[0267] 4.2. Somatic mutations of genes in malignancies with only a single genetic cause, such as retinoblastoma caused by biallelic inactivation of RB1. Code name OP2.

[0268] 4.3. Located in one of the hotspots in cancerhotspots.org with a specific amino acid change count in cancerhotspots.org of less than 10. If the somatic variant is in a tumor type that is not adequately covered in cancerhotspots.org, resources such as COSMIC or tumor type-specific studies may be used. Code OP3.

[0269] 4.4. In the Genome Aggregation Database (gnomAD), the variant does not exist in the reference genome. Code name OP4.

[0270] 5) The strength of the benign evidence is “very strong” (-8 points), including:

[0271] 5.1. In 5 common continental populations (Africans, East Asians, Europeans (non-Finns), Latin Americans, and South Asians), the minor allele frequency in the Genome Aggregation Database (gnomAD) is greater than

[0272] 5%. Codename BVS1. However, if the somatic variant is in a gene known to confer an inherited cancer susceptibility, the ACMG / AMP ClinGen Germline Panel gene-specific guidelines must be consulted to establish the threshold for considering disease prevalence.

[0273] 6) The strength of the benign evidence is “strong” (-4 points), which includes:

[0274] 6.1. In in vitro or in vivo functional studies, it has been well confirmed that the variant has no carcinogenic effect and is a non-pathogenic variant. Code BS1.

[0275] 6.2. In five common continental populations (Africans, East Asians, Europeans (non-Finns), Latin Americans, and South Asians), the minor allele frequency in the Genome Aggregation Database (gnomAD) is greater than

[0276] 1%. Codename BS2. However, if the somatic variant is located in a gene known to cause hereditary cancer susceptibility

[0277] In this setting, the ACMG / AMP ClinGen Germline Panel gene-specific guidelines (if available) must be consulted to establish thresholds for considering disease prevalence.

[0278] 7) The strength of the benign evidence is “supporting” (-1 point), which includes:

[0279] 7.1. Multiple lines of evidence from computational algorithms suggest that the variant has no effect, including conservation / evolution and effects on splicing

[0280] No matter how many algorithms are used for evaluation, this rule can only be calculated once. Code BP1.

[0281] 7.2. A synonymous (silent) variant, which the splicing prediction algorithm predicts has no effect on the splicing consensus sequence, will not generate a new splicing site, and the nucleotide is not highly conserved. Code name BP2.

[0282] After determining which strengths of evidence a variant can obtain, use the following table to obtain the scores for each piece of evidence:

[0283] Strength of evidence score

[0284]

[0285]

[0286] Add up all the scores of the evidence to get the total score of the variant site. Then determine the pathogenicity / benignity classification of the variant according to the score range in the following table:

[0287] Score range for somatic mutation classification

[0288]

[0289] The following calculation is based on the BRAF V600E mutation as an example:

[0290] 1) In the past, in vitro and in vivo functional studies have fully demonstrated that the variant has a significant oncogenic effect, so it meets the OS2 rule.

[0291] 2) In the cancerhotspots.org database, a total of 897 BRAF V600 mutations were recorded, of which 833 were V600E mutations, thus meeting the OS3 rule.

[0292] 3) It is known that the V600E mutation is located in the key functional domain of the BRAF protein, "PTK protein tyrosine kinase", and therefore meets the OM1 rule.

[0293] 4) Calculations using the prediction software RAVEL and SpliceAI both support the carcinogenic effect of BRAF V600E mutation, thus meeting the rules of OP1.

[0294] 5) In the Genome Aggregation Database (gnomAD), the variant does not exist in the reference genome and therefore meets the rules of OP4.

[0295] 6) The total score of the above items is calculated as OS2+OS3+OM1+OP1+OP4=4+4+2+1+1=12.

[0296] 7) According to the score range, the variant is considered "pathogenic (P)".

[0297] Multiple variants are usually detected in a sample, and the classification of the variant with the highest score (that is, the direction that indicates pathogenicity to the greatest extent) is used as the overall variant detection classification of the sample. For example, if BRAF V600E mutation and STK11 A1047T mutation are detected in a sample at the same time, the former is classified as pathogenic (P) and the latter is classified as likely benign (LB), then the overall variant detection classification of the sample is pathogenic (P).

[0298] The fusion gene variation detection results do not refer to the above interpretation process. Each sample detected with a fusion gene variation within the detection range of the present invention is classified as pathogenic (P).

[0299] 13. Final judgment on the benign or malignant nature of the tested samples. The specific judgment process is as follows:

[0300] a. For the three types of mutations including BRAF V600E, TERT promoter mutation, and RET fusion gene, which are clearly highly malignant mutations, the interpretation is definitely pathogenic (P), that is, malignant. There is no need to interpret the RNA expression analysis results.

[0301] b. For samples containing other types of mutations, if at least one of the following two conditions is met, it is judged to be malignant:

[0302] i. Based on the AMP / ASCO / CAP guidelines for the interpretation of somatic variants, the variant is judged to be pathogenic (P) or likely pathogenic (LP);

[0303] ii. Based on RNA expression profiling, it is determined to be malignant.

[0304] 14. The qualitative results are submitted to clinicians for judgment to guide subsequent diagnosis and disease management decision-making. In addition to referring to the overall variant detection classification given in the present invention, clinicians will also study other clinical interpretation information of all variants, and make a diagnosis conclusion and subsequent clinical decision after comprehensive evaluation.

[0305] In a real-world cohort study, 430 Bethesda III-V samples were collected according to natural distribution; based on the postoperative pathological results as the gold standard, 325 of them were malignant samples. The samples were analyzed using the detection and interpretation method of the present invention, and the confusion matrix of the obtained results is shown in the following table:

[0306]

[0307] After the benign and malignant prediction is performed based on the aforementioned process of the present invention, the calculated sensitivity is 96.9%, the specificity is 76.2%, and the overall accuracy is 91.8%. At the same time, the detection sensitivity of the traditional BRAFV600E, TERT promoter, and RET fusion three gene mutation targets in this cohort is only 70.5%. Therefore, the detection method described in the present invention can greatly improve the sensitivity while maintaining effective specificity, and achieve the clinical expected effect of the high-sensitivity Rule-Out of the present invention (that is, ensuring a higher accuracy of benign detection).

[0308] The above examples are for the purpose of illustrating the embodiments disclosed by the present invention and are not to be construed as limiting the present invention. In addition, the various modifications listed herein and the variations of the methods in the invention are obvious to those skilled in the art without departing from the scope and spirit of the present invention. Although the present invention has been specifically described in conjunction with various specific preferred embodiments of the present invention, it should be understood that the present invention should not be limited to these specific embodiments. In fact, various modifications as described above that are obvious to those skilled in the art to obtain the invention should be included within the scope of the present invention.

Claims

1. An auxiliary diagnosis system for benign and malignant thyroid nodules based on the combination of genomic variation and abnormal expression, comprising the following modules: DNA data module: used to obtain the data of the sequencing library constructed by shingled multiple amplification of the DNA of each sample; RNA data module: used to obtain the data of constructing sequencing libraries after multiple amplification of RNA of each sample after reverse transcription and the RNA expression level data of the target genome of each sample; The target genome includes: 14 characteristic genes and 7 internal reference genes; the 14 characteristic genes are AGGF1, BANP, CDK12, CITED1, DICER1, ERBB3, ERBB4, FGFR2, FGFR4, IGF2BP3, KIT, MET, VHL, and ZBTB8A; The first analysis module is used to perform bioinformatics analysis on the data of the sequencing libraries obtained in the DNA acquisition module and the RNA acquisition module, and obtain the benign or malignant judgment results of the thyroid nodules of each sample in combination with the somatic cell variation interpretation guide; The second analysis module is used to process and analyze the RNA expression level data of the target genome of each sample, and to predict the benign and malignant nature of the thyroid nodules for each sample based on the constructed benign and malignant thyroid nodule prediction model to obtain the benign and malignant thyroid nodule prediction results of the corresponding samples; Judgment module: used to judge the final benign or malignant judgment result of each sample based on the benign or malignant judgment result of each sample and the benign or malignant prediction result of each sample; Among them, in the first analysis module, the method of performing bioinformatics analysis on the data of the sequencing library obtained in the DNA data module and the RNA data module includes: performing data quality control on the data of the sequencing library corresponding to the DNA of each sample and the data of the sequencing library corresponding to the RNA; comparing the data of the sequencing library corresponding to the DNA and the data of the sequencing library corresponding to the RNA in each sample that have undergone data quality control with the DNA reference genome and the DNA reference genome to obtain the DNA gene comparison results and the RNA gene comparison results; performing variation detection and fusion detection based on the DNA gene comparison results and the RNA gene comparison results of each sample, and filtering the variation sites; annotating the variation sites after filtering of each sample as detection sites based on the annotation database, and marking their molecular biology and population information characteristics; In the second analysis module, the method of constructing a benign or malignant prediction model for thyroid nodules includes: obtaining RNA expression level data of each characteristic gene and each internal reference gene in multiple benign nodule samples and multiple corresponding malignant papillary thyroid carcinoma samples without BRAF V600E, TERT promoter mutation, and RET fusion; performing processing and analysis based on the RNA expression level data of the target genome of each sample to obtain processing and analysis results of multiple training samples; calculating the normal distribution of each characteristic gene based on the processing and analysis results of each training sample, and constructing a benign or malignant prediction model for thyroid nodules in combination with the benign or malignant results of thyroid nodules of each training sample; Calculating the normal distribution of each characteristic gene based on the processing and analysis results of each training sample includes: obtaining the RPKM value median of each characteristic gene based on the RPKM value ratio of each benign nodule sample in each training sample; calculating the log2 foldchange value of each characteristic gene based on the RPKM value ratio of each characteristic gene of each benign nodule sample and the corresponding RPKM value median, and obtaining the normal distribution of each characteristic gene based on the median of the log2 fold change value of each characteristic gene and the absolute median difference; In the DNA data module, DNA comes from AKT1, ALK, APC, ATM, BANP, BRAF, CDK12, CDKN2A, CHEK2, CTNNB1, DICER1, EGFR, EIF1AX, EP300, ERBB4, EZH1, FAM193A, FARSB, FGFR1, FGFR2, FGFR3, FGFR4, FLT3, GLIS3, GNAQ, GNAS, HRAS, IDH1, IDH2, KIT, KLK1, KM T2C, KMT2D, KRAS, LRP1B, MEN1, MET, MTOR, NCOR2, NF1, NF2, NOTCH1, NRAS, NTRK1, NTRK3, PIK3CA, PLEKHS1, POR, PPARG, PTEN, PTH, RB1, RBM10, RET, ROS1, SMAD4, SPOP, STK11, SUGCT, TP53, TRIM61, TSC2, TSHR, VHL, ZNF148, and TERT genes; In the RNA data module, the RNA comes from a fusion gene; the fusion gene is formed by the fusion of an upstream segment of a first gene breakpoint and a downstream segment of a second gene breakpoint; The first gene and its breakpoints are selected from the following: The second gene and its breakpoints are selected from the following:

2. The system according to claim 1, It is characterized in that The sample is selected from a fresh tissue sample of a thyroid nodule, a puncture biopsy cell sample or a paraffin-embedded fixed tissue sample.

3. The system according to claim 1, It is characterized in that In the DNA data module, the coverage of the shingled multiplex amplification is the entire exon region of the DNA gene and the 20-base region extending outward.

4. The system according to claim 3, It is characterized in that In the DNA data module, when the DNA detection object is the TERT gene, the coverage of the shingled multiplex amplification also includes its promoter region.

5. The system according to claim 1, It is characterized in that In the RNA data module, the scope of reverse transcription amplification is the expressed transcripts of the fusion gene.

6. The system of claim 1, It is characterized in that In the DNA data module or RNA data module, the primers used for amplification are designed using NGS-PrimerPlex software; And / or, in the DNA data module or the RNA data module, the ratio of primers used for amplification is determined by collecting primer feature data and optimizing based on a vector regression model; and / or, in the DNA data module or the RNA data module, when constructing the sequencing library, the target fragment is connected to the index for sequencing; And / or, in the first analysis module, the somatic variation interpretation guidelines are selected from the somatic variation interpretation guidelines developed by the AMP / ASCO / CAP organizations in the United States.

7. The system of claim 6, It is characterized in that The primer characteristic data includes one or more combinations of primer length and GC content, amplification region length and GC content, 5' end and 3' end ΔG, and similarity to other regions of the genome; And / or, the index is selected from a single-end index and a dual-end index; the index is UDI; the index also includes UMI; And / or, the sequencing adapter is selected from the universal sequencing adapter of Illumina or the sequencing adapter of BGI; And / or, the sequencing platform is selected from Illumina's universal sequencing platform or BGI's sequencing platform; And / or, the sequencing method is selected from reversible terminal termination sequencing or combined probe anchor polymerization sequencing.

8. The system of claim 1, It is characterized in that Combined with the somatic cell variation interpretation guidelines, the results of the benign and malignant judgment of thyroid nodules include: Based on the somatic variation interpretation guidelines, the detection sites are interpreted to obtain the benign and malignant judgment results of thyroid nodules; among them, the types of benign and malignant judgment results of thyroid nodules include: pathogenic, possibly pathogenic, unknown significance, possibly benign, and benign.

9. The system of claim 8, It is characterized in that If the detection site of the sample corresponds to one or more of the three types of mutations including BRAF V600E, TERT promoter mutation and RET fusion gene, it will be interpreted as pathogenic based on the somatic mutation interpretation guidelines.

10. The system of claim 1, It is characterized in that The internal reference genes are selected from the expression profile data of cancer tissue and adjacent tissue of thyroid cancer patients in the TCGA database; the 7 internal reference genes are SLC5A5, B2M, PPIH, HPRT1, NONO, PUM1, and SDHA.

11. The system of claim 1, It is characterized in that The processing and analysis of RNA expression level data of the target genome of each sample includes: Data quality control was performed on the RNA expression level data of each sample; The RNA expression level data of each sample that has passed data quality control are sequenced with the corresponding reference genome to obtain the number of reads of the corresponding gene amplicon of each gene in the target genome of each sample; Based on the number of reads of the corresponding gene amplicon of each gene in the target genome of each sample, the RPKM value of each gene in the target genome of each sample was calculated; Calculate the ratio between the RPKM value of each characteristic gene in each sample and the total RPKM value of the internal reference gene to obtain the RPKM value ratio corresponding to each characteristic gene in each sample; Based on the K-means clustering method, outlier samples were determined based on the RPKM value ratios corresponding to each characteristic gene in each sample, and the outlier samples were excluded.

12. The system according to claim 11, It is characterized in that Methods for predicting benign or malignant thyroid nodules for each sample include: The log2 fold change value of each characteristic gene was calculated by comparing the RPKM value ratio corresponding to each characteristic gene of each sample with the corresponding RPKM value median; Based on the normal distribution of each characteristic gene, the gene expression variation results of each characteristic gene of each sample were obtained according to the calculated log2 fold change value of each characteristic gene; Based on the gene expression variation results of each characteristic gene of each sample, the prediction results of benign and malignant thyroid nodules corresponding to each sample are obtained.

13. The system of claim 12, It is characterized in that Based on the normal distribution of each characteristic gene, the method of obtaining the gene expression variation result of each characteristic gene of each sample according to the calculated log2 fold change value of each characteristic gene includes: Based on the gene expression variation result judgment rule, determine whether the calculated log2 fold change value of each characteristic gene is within the 95% confidence interval of the normal distribution of the corresponding characteristic gene, and obtain the gene expression variation result of each characteristic gene of each sample; Among them, the rules for judging gene expression variation results include: If it is within the 95% confidence interval of the normal distribution, it is judged that the corresponding characteristic gene has not mutated; If it is outside the left side of the 95% confidence interval of the normal distribution, the corresponding characteristic gene is judged to be a low-expression variation; If it is outside the right side of the 95% confidence interval of the normal distribution, the corresponding characteristic gene is judged to be overexpressed.

14. The system of claim 1, It is characterized in that Methods for judging the final benign or malignant thyroid nodule judgment result of each sample based on the benign or malignant thyroid nodule judgment result of each sample and the benign or malignant thyroid nodule prediction result include: If the benign or malignant judgment result of the thyroid nodule of the sample corresponds to pathogenic or possibly pathogenic, the final benign or malignant judgment result of the corresponding sample is judged to be malignant; If the benign or malignant judgment result of the thyroid nodule of the sample is unclear, possibly benign, or benign, the final benign or malignant judgment result of the thyroid nodule is obtained based on the benign or malignant prediction result of the thyroid nodule of each sample.

Citation Information

Patent Citations

  • Multi-PCR amplification optimization method, system and equipment based on machine learning and medium

    CN116092585A

  • Primer for detecting variation of benign and malignant related genes of thyroid nodules, kit and detection method

    CN110241215A

  • Thyroid nodule benign and malignant detection method, kit and application

    CN116219016A

  • Mirnas as biomarkers for distinguishing benign from malignant thyroid neoplasms

    US20120157334A1

  • A new method for diagnosing a thyroid tumour and kit thereof

    WO2021116938A1