Construction and application of cancer prediction model

Abstract

The invention provides construction and application of a cancer prediction model. Specifically, the present invention develops a model for cancer prediction, including cancer prediction of whether a cancer is present, and develops a cancer prediction system based on the cancer prediction model. The cancer prediction system provided by the invention has an auxiliary diagnosis result for efficiently and accurately judging whether the tissue to be detected is cancer or not.

Description

Technical Field

[0001] The present invention relates to the field of cancer detection, and in particular, to the construction and application of a cancer prediction model. Background Art

[0002] The development process of malignant tumors includes the development from normal tissue to precancerous lesions and finally to malignant tumors. The precancerous lesions of most malignant tumors are long and may be reversible. Early diagnosis of precancerous lesions in clinical practice is of great significance for disease intervention. On the one hand, it can prevent the progression of the disease in patients, and on the other hand, it can improve the quality of life of patients. With the continuous development and progress of clinical diagnostic technology, there are currently many diagnostic methods, such as pathogen biology examination, cytology, endoscopy, biopsy pathological diagnosis and genetic testing; these technologies have significantly improved the early diagnosis rate of precancerous lesions.

[0003] However, these diagnostic methods still have a certain rate of missed diagnosis and misdiagnosis. At present, pathological diagnosis is the gold standard for most tumor precancerous lesions, but the number of cells in precancerous lesions is relatively small and scattered, and the cell morphology may be similar to normal cells or similar to cells in benign lesions. Therefore, it is difficult to detect and diagnose precancerous lesion cells in tissue specimens.

[0004] Therefore, there is an urgent need in this field to develop a new, more accurate and efficient cancer auxiliary diagnosis method or device. Summary of the invention

[0005] The present invention provides a new cancer auxiliary diagnosis method and device.

[0006] In a first aspect of the present invention, a method for constructing a cancer prediction model is provided, comprising the steps of:

[0007] (s1) Process the third-generation sequencing data and RNA sequencing (RNA-seq) data of cancer tissue samples, paracancerous tissue samples, and normal tissue samples to obtain reference GTF files of full-length RNA transcripts and tumor-specific RNA transcripts (Tu-SRT, Tumor Specific RNA Transcript);

[0008] (s2) providing training set, test set and validation set data for model construction, wherein the data includes expression information (or quantitative results) of full-length RNA transcripts of RNA-seq data obtained according to the reference GTF file after RNA sequencing; the expression information includes expression information of the tumor-specific RNA transcripts;

[0009] Wherein, the expression information of the tumor-specific RNA transcripts includes the expression information of the tumor-specific RNA transcripts from the cancer-positive samples and the expression information of the tumor-specific RNA transcripts from the cancer-negative samples;

[0010] (s3) using the training set data to train a machine learning model selected from the group consisting of a random forest model, a decision tree model, an XGBoost model, a LightGBM model, and a CatBoost model; thereby obtaining hyperparameters of the machine learning models, respectively, and obtaining a trained machine learning model; and

[0011] (s4) Using the test set and validation set data, the trained machine learning model is scored for the prediction results of the machine learning model using the following scoring indicators: AUC value, accuracy, sensitivity and specificity; the machine learning model whose score meets the expected value is used as a cancer prediction model.

[0012] In another preferred embodiment, the expression information of the transcript is the TPM (Transcripts Per Million) value of the transcript expression.

[0013] In another preferred embodiment, the validation set data and the training set and test set data are derived from different cohort samples.

[0014] In another preferred example, the machine learning model is a LightGBM model.

[0015] In another preferred embodiment, the expression information of the transcript is represented by a TPM (Transcript per million) value.

[0016] In another preferred embodiment, in step (s3), the training set is used to train a machine learning model selected from the following group: a random forest model, a decision tree model, an XGBoost model, a LightGBM model, and a CatBoost model; thereby obtaining the hyperparameters of the five machine learning models respectively, and obtaining five trained machine learning models;

[0017] Furthermore, in step (s4), the five trained machine learning models are scored using the test set and the validation set, using the following scoring indicators: AUC value, accuracy, sensitivity and specificity; and the machine learning model with the highest score is used as the cancer prediction model.

[0018] In another preferred embodiment, the cancer positive sample is a cancer tissue sample.

[0019] In another preferred embodiment, the cancer sample includes a cancer tissue sample, a cancer cell sample, or a combination thereof.

[0020] In another preferred embodiment, the cancer samples include samples of n different cancers, wherein n is a positive integer ≥10.

[0021] In another preferred embodiment, n≥15; preferably, n is 20-100; more preferably, n is 20-50.

[0022] In another preferred embodiment, the cancer is selected from the group consisting of adrenocortical carcinoma, bladder urothelial carcinoma, invasive breast cancer, cervical squamous cell carcinoma and adenocarcinoma, bile duct cancer, colon cancer, diffuse large B-cell lymphoma, esophageal cancer, multiforme glioma, head and neck squamous cell carcinoma, renal chromophobe cell carcinoma, renal clear cell carcinoma, renal papillary cell carcinoma, acute myeloid leukemia, brain low-grade glioma, hepatocellular carcinoma, lung adenocarcinoma, lung squamous cell carcinoma, mesothelioma, ovarian serous cystadenocarcinoma, pancreatic cancer, pheochromocytoma and paraganglioma, prostate cancer, rectal adenocarcinoma, sarcoma, skin melanoma, gastric cancer, testicular cancer, thyroid cancer, thymic cancer, endometrial cancer, uterine sarcoma, uveal melanoma, or a combination thereof.

[0023] In another preferred embodiment, the cancer-negative sample is a paracancerous tissue sample and / or a normal tissue sample.

[0024] In another preferred embodiment, in step (s1), it includes: establishing a full-length transcript reference GTF file by processing third-generation sequencing data and filtering transcript annotations; the GTF file is used to quantify RNAseq transcripts obtained by sample RNA sequencing; and

[0025] Compare the RNAseq transcript expression information of different tissue samples, obtain the types of tumor-specific RNA transcripts, and obtain the expression information of each tumor-specific RNA transcript in each cancer sample.

[0026] In another preferred example, the expression information of the transcript includes the expression information of the transcript.

[0027] In another preferred embodiment, in step (s1), the following steps are included:

[0028] (a) Collect the third-generation sequencing data and RNA-seq data of cancer tissues and adjacent tissue samples from different tumor sources, as well as normal tissue samples, and generate non-redundant transcriptome GTF files;

[0029] (b) optimizing the non-redundant transcriptome GTF file, wherein, for a certain cancer / tissue RNA-seq, if all splicing sites of a transcript are detected in ≥m samples, the transcript is retained, where m is 1-90%, preferably 2-50%;

[0030] (c) Each transcript was compared with the reference transcriptome (GENCODE v.35), and the transcripts were mainly divided into four groups, including full splicing matches, incomplete splicing matches, novel in the catalog, and novel out of the catalog. After filtering, a reference GTF file for full-length RNA transcripts was established:

[0031] Among them, for all incomplete splicing matching transcripts, such transcripts are filtered out from the entire transcriptome;

[0032] Filter out unqualified novel transcripts in the catalog and outside the catalog (for example, cage peaks and polyA sites were annotated using SQANTI3, and novel transcripts in the catalog and outside the catalog were filtered when there were 16 or more adenines within 20 bases downstream of the annotated transcription termination site);

[0033] For similar transcripts, the longest transcript is retained and other transcripts are filtered out;

[0034] (d) Based on the reference GTF file for the full-length RNA transcript, the full-length transcript is quantified on the tissue sample RNA-seq data, thereby obtaining the full-length RNA transcript expression information of the tissue sample.

[0035] (e) obtaining, by comparison, RNA transcripts having significantly different expression information between the cancer tissue sample and adjacent tissue samples and normal tissue samples, thereby obtaining the tumor-specific RNA transcripts.

[0036] In another preferred embodiment, the significant difference is that the gene is expressed in tumor tissue but not expressed or very low expressed in adjacent adjacent tissue or normal tissue.

[0037] In another preferred embodiment, step (d) includes:

[0038] Salmon software was used to align the RNA-seq short read sequences obtained by RNA-seq sequencing to the transcript library corresponding to the GTF file, and the expression level of the transcripts in each sample was measured in the form of TPM (Transcript per million) to obtain the transcript expression information of the sample.

[0039] In another preferred embodiment, the types of tumor-specific RNA transcripts are obtained by the following steps:

[0040] (a) obtaining sequencing data of biopsy tissue samples from the population to be tested (including sequencing including third-generation sequencing, second-generation sequencing or a combination thereof);

[0041] (b) by processing the third-generation sequencing data and filtering the transcript annotations, a full-length transcript reference GTF file is established; the GTF file is used to quantify the RNAseq transcripts obtained by the second-generation sequencing of tissue samples;

[0042] (c) The RNA-seq short-read sequence is aligned to the transcript library obtained by long-read sequencing, and the expression level of each transcript in each sample is measured in the form of TPM (Transcript per million) to generate transcript expression information of the tissue to be tested; the transcript expression information includes the expression information of tumor-specific RNA transcripts of the tissue to be tested.

[0043] In a second aspect of the present invention, a reference GTF file for a full-length RNA transcript is provided, obtained by the following steps:

[0044] (i) Collect the third-generation sequencing data and RNA-seq sequencing data of cancer tissues and adjacent tissue samples from different tumor sources, as well as normal tissue samples, and generate non-redundant transcriptome GTF files;

[0045] (ii) optimizing the non-redundant transcriptome GTF file, wherein, for a certain cancer / tissue RNA-seq, if all splicing sites of a transcript are detected in ≥m samples, the transcript is retained, where m is 1-90%, preferably 2-50%;

[0046] (iii) Each transcript was compared with the reference transcriptome (GENCODE v.35), and the transcripts were mainly divided into four groups, including full splicing matches, incomplete splicing matches, novel in the catalog, and novel out of the catalog. After filtering, a reference GTF file for full-length RNA transcripts was established:

[0047] Among them, for all incomplete splicing matching transcripts, such transcripts are filtered out from the entire transcriptome;

[0048] Filter out unqualified novel transcripts in the catalog and outside the catalog (for example, cage peaks and polyA sites were annotated using SQANTI3, and novel transcripts in the catalog and outside the catalog were filtered when there were 16 or more adenines within 20 bases downstream of the annotated transcription termination site);

[0049] For similar transcripts, the longest transcript is retained and other transcripts are filtered out.

[0050] In another preferred embodiment, the full splicing match, incomplete splicing match, novel in the catalog and novel out of the catalog are, respectively, transcripts that are completely consistent with known isoforms in the reference transcriptome database compared to the reference transcriptome; transcripts that are completely contained by known transcripts but have fewer exons than known transcripts, and whose intron sequences completely correspond to known transcripts; transcripts that display known splicing sites but present new splicing patterns or exon combinations; and transcripts that have at least one splicing site that has not been annotated in existing databases.

[0051] In a third aspect of the present invention, a method for determining expression information of a full-length RNA transcript is provided, comprising the steps of:

[0052] (i) Based on the reference GTF file for full-length RNA transcripts described in the second aspect of the present invention, full-length transcripts are quantified on tissue sample RNA-seq data to obtain full-length RNA transcript expression information.

[0053] In another preferred example, the full-length RNA transcript expression information includes expression information of the full-length RNA transcript expression spectrum.

[0054] In a fourth aspect of the present invention, a cancer prediction system is provided, the system comprising:

[0055] (a) an input module, wherein the input module is configured to input the full-length RNA transcript expression information of the sample to be tested;

[0056] (b) an evaluation module (or prediction module): the evaluation module is configured to input the full-length RNA transcript expression information into the cancer prediction model constructed by the method described in the first aspect of the present invention, thereby obtaining a cancer prediction result of the sample to be tested; and

[0057] (c) An output module, wherein the output module is configured to output the prediction result.

[0058] In another preferred example, the system further includes (a0) a preprocessing module, which is configured to analyze the RNA-seq sequencing information of the sample to be tested using the method described in the third aspect of the present invention to determine the full-length RNA transcript expression information of the sample to be tested.

[0059] In another preferred example, the system further includes (d) a control module, wherein the control module is configured to control the operation of each module.

[0060] In another preferred example, the system further includes (e) a storage module, which is configured to store the following data: hyperparameters of the model, predicted probabilities, and preset reference thresholds.

[0061] In another preferred embodiment, the determination in the prediction module includes:

[0062] The full-length RNA transcript expression information is input into the cancer prediction model to obtain a prediction result, which includes a prediction probability; when the prediction probability in the prediction result is ≥ a preset reference threshold for cancer prediction, it is judged to be positive, that is, the sample to be tested is a cancer sample; otherwise, it is judged to be negative, that is, the sample to be tested is a non-cancer sample.

[0063] In a fifth aspect of the present invention, a computer storage medium is provided, wherein the storage medium is used to store a computer program corresponding to the algorithm of the cancer prediction model constructed by the method described in the first aspect of the present invention.

[0064] In a sixth aspect of the present invention, a method for auxiliary diagnosis of cancer is provided, comprising the steps of:

[0065] (a) Providing data: providing RNA sequencing information of the sample to be tested;

[0066] (b) Data preprocessing: for the sequencing information, analyzing and determining the full-length RNA transcript expression information of the sample to be tested, or using the method described in the third aspect of the present invention to analyze and determine the full-length RNA transcript expression information of the sample to be tested;

[0067] (c) analysis and evaluation: inputting the full-length RNA transcript expression information into the cancer prediction model constructed by the method described in the first aspect of the present invention, thereby obtaining an evaluation result of the sample to be tested; and

[0068] (d) Output results: Output the evaluation results.

[0069] In another preferred embodiment, the sample to be tested is selected from the following group: a cell sample, a tissue sample, or a combination thereof.

[0070] In another preferred embodiment, the method is an in vitro method.

[0071] In another preferred embodiment, the method is non-diagnostic and non-therapeutic.

[0072] It should be understood that within the scope of the present invention, the above-mentioned technical features of the present invention and the technical features specifically described below (such as embodiments) can be combined with each other to form a new or preferred technical solution. Due to space limitations, they will not be described one by one here. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] Figure 1 A schematic diagram of the cancer prediction model in Example 2 is shown; Feat in the figure refers to feature.

[0074] Figure 2 The ROC curves of the cancer prediction model generated by LightGBM model training in the training set, test set, and validation set are shown.

[0075] Figure 3 A schematic diagram of a prediction module according to an example of the present invention is shown.

[0076] Figure 4 A schematic diagram of a process for predicting cancer in a population to be detected according to an example of the present invention is shown. DETAILED DESCRIPTION

[0077] The inventors have conducted extensive and in-depth research, based on the results of the third generation sequencing and the second generation sequencing (including RNA-seq) of tumor tissue and paracancerous tissue and / or normal tissue samples, and quantified the RNA transcripts to determine the expression information of the full-length RNA transcripts of the tissue samples, and based on this, have discovered tumor-specific RNA transcripts. And based on the tumor-specific RNA transcripts, a model for cancer prediction has been developed, and a system (or device) for cancer prediction has been developed based on the model. The cancer prediction model or system of the present invention has a higher prediction accuracy. It can be used for early clinical auxiliary diagnosis. On this basis, the present invention has been completed.

[0078] the term

[0079] In order to more easily understand the present disclosure, some terms are first defined. As used in this application, unless otherwise expressly provided herein, each of the following terms should have the meaning given below. Other definitions are set forth throughout the application.

[0080] The term "about" can refer to a value or composition that is within an acceptable error range for a particular value or composition as determined by one of ordinary skill in the art, which will depend in part on how the value or composition is measured or determined. For example, as used herein, the expression "about 100" includes all values ​​between 99 and 101 (e.g., 99.1, 99.2, 99.3, 99.4, etc.).

[0081] As used herein, the term "comprising" or "including (comprising)" may be open, semi-closed and closed. In other words, the term also includes "consisting essentially of" or "consisting of".

[0082] As used herein, unless otherwise indicated, any concentration range, percentage range, ratio range or integer range should be understood to include any integer value within the range and fractional values ​​thereof (e.g., one tenth and one hundredth of an integer) where appropriate.

[0083] As used herein, the term "and / or" refers to and encompasses any and all possible combinations of one or more of the associated listed items.

[0084] As used herein, "transcript expression information", "transcript quantitative results", "transcript expression level" and "transcript expression value" have the same meaning and can be used interchangeably, all referring to the alignment of the RNA-seq short read sequence of the tissue sample to the transcript library corresponding to the GTF file constructed by the present invention, and the expression level of each transcript in the sample measured in the form of TPM (Transcript per million).

[0085] As used herein, "full-length RNA transcript" and "full-length transcript" have the same meaning and can be used interchangeably, and both refer to the complete RNA molecule produced from a gene, including all parts of the coding region and non-coding region. It contains a series of information such as the 5' untranslated region (5'UTR), exon region, intron region (which is cut out in mature mRNA but retained in the full-length transcript) and 3' untranslated region (3'UTR).

[0086] Tumor-specific RNA transcripts

[0087] Tumor-specific RNA transcripts (Tumor-SRT or Tu-SRT) are defined as being found only in tumor samples and not in normal samples (or at very low levels in normal samples). The SRTs of different types of tumors determined by RNA-seq are based on the following criteria:

[0088] i) Transcripts with a TPM>0.5 in >5% of samples in a certain tumor type were considered to be expressed in that tumor type, and the median value of samples with TPM>0.5 was used as its expression level;

[0089] ii) The expression of transcripts in normal tissues is divided into two cases. If the number of TPM>0.5 observed in all normal tissues (except testis) in the GTEx dataset and adjacent tissues in the TCGA dataset is not less than 3, the maximum value is used to represent the expression. If the number of TPM>0.5 observed is less than 3, the median value is used to represent the expression;

[0090] iii) The transcript is expressed in the tumor at a level at least 10-fold higher than the maximum expression level observed in all normal tissues (except testis) and adjacent adjacent tissues.

[0091] According to the above calculation method, the present invention identified 44673 tumor tissue-specific RNA transcripts. These specific RNA transcripts are only highly expressed in tumors, and are not expressed or very low expressed in normal tissues and adjacent tissues (the above list of 44673 tumor tissue-specific RNA transcripts is calculated by the inventors and is unique to the present invention).

[0092] Tissue sample full-length RNA transcript quantification method of the present invention

[0093] The method for quantifying full-length RNA transcripts in tissue samples of the present invention comprises the following steps:

[0094] (St1): 177 tumor tissue specimens from patients with hepatocellular carcinoma, colorectal cancer, ovarian cancer, breast cancer, nasopharyngeal carcinoma, neuroendocrine tumors, gastric cancer, gastrointestinal stromal tumors, renal cancer and cervical cancer were collected. RNA was extracted from the collected tissues and then subjected to PacBio Iso-seq third-generation sequencing. The original PacBio sequencing data was processed using the Iso-Seq workflow.

[0095] (St2): Collect third-generation PacBio and ONT transcriptome sequencing data of 21 cancer tissue samples (including esophageal cancer, lung cancer, glioblastoma, leukemia, lymphoma, melanoma, myeloma and sarcoma) and 153 normal tissue samples (including adipose tissue, adrenal gland, brain, breast, heart, liver, lung, muscle, ovary, pancreas, testis, fibroblasts, GM12878, H1, H9, HEK293T and WTC11 cells) from public databases.

[0096] (St3): FASTQ sequences from PacBio and ONT data were aligned using minimap2 and processed using the standard pipeline to obtain a merged GTF transcriptome. Full-length transcripts from all samples were merged into a non-redundant transcriptome GTF file using gffcompare software.

[0097] (St4): In the merged GTF file, if all splice sites of a transcript are detected in at least 5 samples in TCGA and GTEx cancer / tissue RNA-seq, the transcript is retained.

[0098] (St5): These transcripts were compared with the reference transcriptome (GENCODEv.35) using SQANTI3 and gffcompare software, and the transcripts were mainly divided into four groups, including full splicing matches, incomplete splicing matches, novel in the catalog, and novel outside the catalog. All incomplete splicing match transcripts may be the result of RNA degradation and incomplete reverse transcription, and such transcripts were filtered out of the entire transcriptome.

[0099] (St6): Cage peaks and polyA sites were annotated using SQANTI3, and novel transcripts both in- and outside the catalog were filtered when there were 16 or more adenines within 20 bases downstream of the annotated transcription stop site.

[0100] (St7): The full-length sequence of the transcript was obtained using gffread software according to the transcript position provided by GTF. ORFs in the transcripts of the third-generation sequencing data were predicted using GeneMark. When the stop codon was located more than 50 nucleotides upstream of the last exon-exon splice site, the transcript was predicted to induce nonsense-mediated decay.

[0101] (St8): The whole genome of human hg38 transposable elements was downloaded from the UCSC genome browser, and the BEDTools software was used to determine whether the transcripts overlapped with TE within TSS ± ​​100bp. Similar transcripts were merged according to the following criteria: i) up to three exon differences were allowed. ii) each exon did not differ by more than four bases. iii) the upper limit of the dissimilarity allowed for transcription start and stop sites was 100. iv) for coding transcripts, it was necessary to ensure that the ORFs between transcripts did not differ.

[0102] (St9): For similar transcripts, the longest one was retained and other transcripts were filtered out. A reference GTF file containing 1,069,895 full-length RNA transcripts was established for full-length transcript quantification of tissue sample RNAseq data.

[0103] (St10): During data processing, Salmon software can be used to create an index for the transcript GTF file containing 1,069,895 full-length RNA transcripts obtained above.

[0104] (St11): The RNA-seq short read sequences were aligned to the transcript library, and the expression levels of transcripts in each sample were measured in the form of TPM (Transcript per million). A quantitative method was developed to quantify the expression profile of the full-length RNA transcripts of samples using RNAseq data.

[0105] Cancer prediction model based on machine learning

[0106] The machine learning-based cancer prediction model of the present invention relies on the common biological characteristics of cancer cells in different tissues (tumor-specific RNA transcripts, Tumor-SRT), uses machine learning to set positive and negative thresholds, and constructs a prediction model. By comparing the specific transcript expression results of cancer tissues with paracancerous tissues and normal tissues, the training set model and the validation set model are continuously optimized and improved to diagnose whether the patient has cancer. The cancer prediction system device uses the cancer prediction model to predict based on the tumor-specific transcript (SRT) data expressed by the patient's tissue to diagnose whether the patient has cancer.

[0107] As a branch of artificial intelligence, machine learning allows computer systems to learn from data and automatically improve by using statistical and computer science methods, and to make predictions and decisions by analyzing and identifying patterns and rules in data. In the field of oncology, it can analyze data such as patients' clinical symptoms, medical images, laboratory tests, and histopathology to discover new patterns and correlations between variables and generate predictive models. By applying machine learning technology, the field of oncology diagnosis can extract valuable information from large amounts of complex data, helping doctors provide more accurate and rapid diagnostic results, and improving the treatment effects and survival rates of cancer patients.

[0108] In the model construction process of the present invention, a machine learning model is used to construct the final model of the present invention. The machine learning model includes: a random forest model, a decision tree model, an XGBoost model, a LightGBM model and a CatBoost model.

[0109] Cancer auxiliary diagnosis of the present invention

[0110] The present invention provides multiple applications of the cancer auxiliary diagnosis model of the present invention, including the development of cancer auxiliary diagnosis software, equipment or system. Figure 4 A schematic diagram of a process for predicting cancer in a population to be tested based on the cancer auxiliary diagnosis model of the present invention is shown.

[0111] like Figure 4 As shown, first obtain the full-length transcript expression information of the object to be tested, the steps include: after obtaining the biopsy tissue sample from the population to be tested, the full-length transcript reference GTF file established by processing the third-generation sequencing data and filtering the transcript annotation is used to quantify the RNAseq transcript of the tissue sample; align the RNA-seq short-read sequence to the transcript library obtained by long-read sequencing, and measure the expression of the transcript in each sample in the form of TPM (Transcript per million), and generate the expression information of the full-length RNA transcript of the tissue to be tested. The transcript expression information includes the tumor-specific RNA transcript of the tissue to be tested.

[0112] After the expression information of the full-length RNA transcript is input into the pre-trained cancer prediction model, the cancer prediction model outputs the prediction value of the tested tissue being a cancer tissue. If the prediction value of the tested tissue being a cancer tissue is greater than the preset prediction threshold, the tested tissue is determined to be positive, and the tested tissue includes cancer tissue, or is a cancer tissue; if it is less than the preset prediction threshold, the tested tissue is determined to be negative. The preset prediction threshold can be set to 50%, and the present invention does not limit the specific size of the preset prediction threshold.

[0113] In another preferred embodiment, the cancer auxiliary diagnosis device, software or system of the present invention includes modules selected from the following groups: an input module, a preprocessing module, a prediction module (or an evaluation module), and an output module.

[0114] In another preferred embodiment, the cancer auxiliary diagnosis system of the present invention makes predictions based on the specific RNA transcript (SRT) data expressed by the patient's tissue to be tested, to diagnose whether the patient has cancer, and the prediction index is objective.

[0115] The technical scheme of the present invention is based on the pre-trained cancer auxiliary diagnosis model, and the probability of whether the current sample tissue is cancer is predicted according to the gene information of the sample tissue, and the gene feature information includes Tu-SRT (tumor-specific transcript). All these feature variables are analyzed by long read length and short read length RNA sequencing technology for tumor and normal tissue samples, and the RNA sequencing three-generation sequencing data from a variety of normal and cancerous human tissues and cells are integrated, and tumor-specific RNA transcripts are analyzed and obtained from a large amount of case data. Five models of random forest model (Random Forest), decision tree model (Decision Tree), XGBoost model, LightGBM model and CatBoost model use grid search to obtain the hyperparameters of each model according to each feature, compare the roc_auc of five models under their hyperparameters, select the model with the highest roc_auc and its hyperparameters, obtain five prediction models of tumors trained, and obtain the optimal model in five prediction models. So as to obtain the probability of the sample to be tested being diagnosed as cancer, as an outcome indicator, it is very objective.

[0116] In addition, the application of this system is independent of pathologists, requires less human intervention, can provide standardization and consistency, and offer pathologists highly stable auxiliary tumor diagnosis, thereby reducing the burden on physicians and improving work efficiency. It has high clinical transformation value and market prospects.

[0117] It should be noted that in the device or apparatus of the present application, the storage medium (computer-readable medium) used to store the model or algorithm of the present invention may be a computer-readable signal medium or a non-transitory computer-readable storage medium or any combination of the above two. The non-transitory computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of non-transitory computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0118] In the present application, a non-transitory computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, apparatus, or device. In the present application, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. A computer-readable signal medium may also be any computer-readable medium other than a non-transitory computer-readable storage medium, which may send, propagate, or transmit a program for use by or in combination with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0119] The computer-readable medium may be included in the electronic device, or may exist independently without being incorporated into the electronic device.

[0120] The main advantages of the present invention include:

[0121] (a) This prediction system can help doctors provide more accurate and rapid cancer diagnosis results, improve diagnostic efficiency, and has good clinical application value and market prospects.

[0122] The present invention will be further described below in conjunction with specific examples. It should be understood that these examples are intended to illustrate the present invention only and are not intended to limit the scope of the present invention. The experimental methods in the following examples where specific conditions are not specified are usually performed under conventional conditions, such as those described in Sambrook et al., Molecular Cloning: A Laboratory Manual (New York: Cold Spring Harbor Laboratory Press, 1989), or under conditions recommended by the manufacturer. Unless otherwise indicated, percentages and parts are weight percentages and weight parts.

[0123] Example 1 Construction of GTF files and screening of tumor-specific RNA transcripts

[0124] 1.1 Data Acquisition

[0125] 177 tumor tissue specimens from patients with hepatocellular carcinoma, colorectal cancer, ovarian cancer, breast cancer, nasopharyngeal carcinoma, neuroendocrine tumors, gastric cancer, gastrointestinal stromal tumors, renal cancer and cervical cancer were collected. RNA was extracted from the collected tissues and PacBio Iso-seq third-generation sequencing and second-generation sequencing (including RNA-seq) were performed (this part of the data was generated by the inventors through sequencing). The original PacBio sequencing data was processed using the Iso-Seq workflow.

[0126] The third-generation PacBio and ONT transcriptome sequencing data and second-generation sequencing data (including RNA-seq) of 21 cancer tissue samples (including esophageal cancer, lung cancer, glioblastoma, leukemia, lymphoma, melanoma, myeloma and sarcoma) and 153 normal tissue samples (including adipose tissue, adrenal gland, brain, breast, heart, liver, lung, muscle, ovary, pancreas, testis, fibroblasts, GM12878, H1, H9, HEK293T and WTC11 cells) from public databases were collected.

[0127] 1.2 Construction of GTF file

[0128] FASTQ sequences from PacBio and ONT data were aligned using minimap2 and processed using a standard pipeline to obtain a merged GTF transcriptome. Full-length transcripts from all samples were merged into a non-redundant transcriptome GTF file using gffcompare software.

[0129] The non-redundant transcriptome GTF files were optimized by various methods to obtain an optimized merged GTF file. In the optimized merged GTF file, if all splicing sites of the transcript were detected in at least 5 samples in TCGA and GTEx cancer / tissue RNA-seq, the transcript was retained.

[0130] These transcripts were compared with the reference transcriptome (GENCODE v.35) using SQANTI3 and gffcompare software, and the transcripts were mainly divided into four groups, including full splicing matches, incomplete splicing matches, novel in the catalog, and novel out of the catalog.

[0131] Among them, fully spliced ​​matching transcripts refer to those that are completely consistent with known isoforms in the database. Incompletely spliced ​​matching transcripts refer to those that are completely contained by known transcripts, but have fewer exons than known transcripts, and their intron sequences completely correspond to known transcripts. Novel transcripts in the catalog show known splicing sites, but present new splicing patterns or exon combinations. Novel transcripts outside the catalog refer to transcripts that have at least one splicing site that is not annotated in existing databases.

[0132] All incompletely spliced ​​transcripts, which may be the result of RNA degradation and incomplete reverse transcription, were filtered out from the entire transcriptome.

[0133] Cage peaks and polyA sites were annotated using SQANTI3, and both novel in- and out-of-catalog transcripts were filtered when there were 16 or more adenines within 20 bases downstream of the annotated transcription stop site.

[0134] The full-length sequence of the transcript was obtained using gffread software according to the transcript position provided by GTF. ORFs in the transcripts of the third-generation sequencing data were predicted using GeneMark. When the stop codon was located more than 50 nucleotides upstream of the last exon-exon splice site, the transcript was predicted to trigger nonsense-mediated decay, and these transcripts were filtered out from the entire transcriptome.

[0135] The whole genome transposable elements of human hg38 were downloaded from the UCSC genome browser, and the BEDTools software was used to determine whether the transcripts overlapped with TE within TSS ± ​​100 bp. Similar transcripts were merged according to the following criteria: i) up to three exon differences were allowed. ii) each exon did not differ by more than four bases. iii) the upper limit of the dissimilarity allowed for the transcription start and end sites was 100. iv) for coding transcripts, it was necessary to ensure that the ORFs between transcripts did not differ.

[0136] For similar transcripts, the longest one was retained and the other transcripts were filtered out, thereby establishing a reference GTF file containing 1,069,895 full-length RNA transcripts (including 1,069,895 transcripts) for full-length transcript quantification of tissue sample RNAseq data (this newly generated GTF file is unique to the inventor after analyzing his own sequencing data and public data).

[0137] During data processing, the Salmon software can be used to create an index for the GTF file containing 1,069,895 full-length RNA transcripts obtained above (creating an index based on a newly generated GTF file is unique to the present invention), thereby establishing a transcript library.

[0138] The RNA-seq short-read sequences of the above samples (data sources include public databases and the second-generation sequencing data obtained by the above self-test) were aligned to the transcript library, and the expression levels of 1,069,895 transcripts in each sample were measured in the form of TPM (Transcript per million). A quantitative method was developed for quantifying the expression profile of the full-length RNA transcripts of samples through RNAseq data.

[0139] 1.3 Acquisition of tumor-specific RNA transcripts

[0140] The inventors developed the FLIBase database based on full-length RNA transcripts by quantifying RNAseq data of 33 TCGA tumors and 28 normal tissues in the public database, creating a comprehensive resource library of full-length transcript information of human cancer tissues and normal tissues. The database covers a variety of cancer types and different normal tissue types, identifies a large number of unannotated tumor-specific RNA transcripts, and provides a panoramic view of tissue-specific RNA transcripts of different normal tissues.

[0141] The FLIBase of the present invention has significant advantages in discovering a large number of previously unannotated isoforms and tumor-specific RNA transcripts. Using this database to conduct in-depth mining and research on the cancer transcriptome at the transcript level will help to obtain new insights into the precise diagnosis and treatment of cancer.

[0142] The tumor-specific RNA transcript (Tumor-SRT) described in the present invention is an RNA transcript that has significant differences in transcript quantification (including TPM value of transcript) between cancer tissue and adjacent tissue and normal tissue, i.e., a transcript that is expressed in cancer tissue but not expressed or very low expressed in adjacent tissue and normal tissue. A total of 44,673 tumor-specific RNA transcripts were obtained in the present invention.

[0143] Example 2 Construction of cancer prediction model

[0144] 2.1 Training set and test set sample data

[0145] Source of sample data: The sample data include 33 cancer tissues (positive) and adjacent cancer tissues (negative) from the TCGA dataset in the public database, and RNAseq data (negative) of normal tissues from the public databases GTEx and GEO, as shown in Tables 1 and 2.

[0146] Table 1 Number of cancer tissue and adjacent tissue samples of 33 cancers in the TCGA dataset in the public database

[0147]

[0148]

[0149] Table 2 Number of normal tissue samples in the public databases GTEx and GEO datasets

[0150]

[0151]

[0152] 2.2 Model construction

[0153] The sample data in 2.1 is divided into training set and test set in a ratio of 3:1. The test set and training set data are independent of each other.

[0154] The expression information or quantitative results (expressed in TPM) of 44,673 tumor-specific RNA transcripts of each sample in the training set were input into the machine learning model for training to obtain a cancer prediction model.

[0155] Hyperparameter determination: Use the expression information of tumor-specific RNA transcripts to perform grid search on the machine learning model to determine the model's hyperparameters. When performing grid search, each parameter combination is cross-validated, such as using 5-fold cross validation and using ROC AUC as the scoring function.

[0156] The machine learning models for training include: Random Forest model, Decision Tree model, XGBoost model, LightGBM model and CatBoost model. Finally, five cancer prediction models trained on five machine learning models were obtained (corresponding to Random Forest model, Decision Tree model, XGBoost model, LightGBM model and CatBoost model, respectively).

[0157] Among them, the cancer tissue samples of the TCGA dataset in the above training set sample data are used as positive samples in the training set, and the adjacent cancer tissue samples and all normal tissue samples in GTEx are used as negative samples in the training set (that is, adjacent cancer tissue samples and all normal tissue samples in GTEx are regarded as normal tissue samples).

[0158] The schematic diagram of the cancer prediction model of this embodiment is as follows Figure 1 As shown. The cancer prediction model can predict cancer and non-cancer, obtain features in tumor-specific RNA transcripts based on a set of cancer feature names, and input them into a cancer classifier. The cancer classifier is pre-constructed with information such as a classifier name and classifier parameters, and the classifier parameters include predetermined hyperparameters. A judgment of cancer or non-cancer is made based on the cancer classifier. If the predicted value output by the cancer classifier is greater than the cancer prediction threshold, it is judged to be cancer; if it is less than the cancer prediction threshold, it is judged to be non-cancer.

[0159] 2.3 Model Validation

[0160] The machine learning for prediction of the five cancers obtained above was verified on the test set and validation set, and measured according to four indicators: AUC value, accuracy, sensitivity, and specificity. The model with the highest score was taken as the cancer prediction model.

[0161] Specifically, the cancer prediction model is evaluated using the test set (described in 2.2) and validation set data, respectively.

[0162] The validation set includes RNA-seq sequencing data of cancer tissues and adjacent tissues of 33 cancers obtained from the public database GEO and dbGAP independent datasets, and the full-length RNA transcript expression information determined according to the GTF file obtained in Example 1, as shown in Table 3.

[0163] Table 3 Cancer tissue and adjacent tissue sample data of 33 cancers in the public database GEO and dbGAP datasets

[0164]

[0165]

[0166] In this validation dataset, all cancer tissue samples in the GEO and dbGAP datasets were used as positive samples, and all adjacent tissue samples were used as negative samples.

[0167] The cancer prediction model was used to score the test set and validation set data respectively, and finally the cancer prediction model was determined to be the cancer prediction model trained according to the LightGBM model (Table 4). The test results or scoring results of the cancer prediction model (or tumor diagnosis model) trained according to the LightGBM model are shown in Tables 5 and Figure 2 shown.

[0168] Table 4 Selection of cancer prediction models

[0169]

[0170] Table 5 Accuracy of cancer prediction model obtained based on LightGBM training

[0171]

[0172] The results show that the accuracy of the test set is 0.9897, the sensitivity is 0.9978, the specificity is 0.9665, and the AUC is 0.9995. The AUC obtained using the validation set is 0.8703, the accuracy is 0.8420, and when the threshold is 0.99, the sensitivity is 0.887 and the specificity is 0.603, which further proves the effectiveness of the model.

[0173] After the above cancer prediction model is established, it is possible to predict whether the subject to be tested has cancer based on the full-length RNA transcript expression information (including tumor SRT) of the subject to be tested and the cancer prediction model.

[0174] Example 3 Cancer auxiliary diagnosis equipment and system

[0175] Based on the cancer auxiliary diagnosis model constructed by the present invention, the present invention develops a cancer auxiliary diagnosis system and software.

[0176] The above-mentioned cancer auxiliary diagnosis system includes, in addition to an input module and an output module, a preprocessing module and a prediction module.

[0177] The preprocessing module is configured to quantify the expression value of the full-length RNA transcript of the sample to be tested.

[0178] Figure 3 A schematic diagram of a prediction module of an example of the present invention is shown. The module extracts the features of the sample to be predicted based on feature data reading and inputs them into a cancer prediction model, and the parameters of the prediction model are pre-set in the model or system. The prediction model predicts the probability of whether the input sample is cancer and its threshold.

[0179] The prediction module is configured to perform the following operations: in response to the received full-length RNA transcript expression information of the tissue to be tested, input it into a pre-trained cancer prediction model to obtain a prediction result.

[0180] Taking cancer as an example, after the expression information of the full-length RNA transcript of the tissue to be tested is input into the cancer prediction model, a candidate prediction result is obtained, including a prediction probability and a preset reference threshold; when the prediction probability in the prediction result is greater than or equal to the preset probability threshold, it is judged as positive, which is used to characterize the target sample as cancer.

[0181] Figure 3 A schematic diagram showing whether a cancer is present or not is shown according to an example of a cancer prediction system of the present invention.

[0182] When predicting cancer, the program generates a cancer test set, reads the optimal Cancer classifier and its hyperparameters according to the configuration file, and makes predictions. It will output the probability that the sample is cancer positive, recorded as Cancer_prob. At the same time, the module uses the trained classifier to predict the training set and calculates the threshold that maximizes the prediction result f1_score of the training set, recorded as threshold. Cancer_prob ≥ threshold is recorded as positive, otherwise it is negative.

[0183] The following is relevant information for the development system or device:

[0184] System name: Specific RNA transcript tumor diagnosis system based on machine learning

[0185] Development hardware environment

[0186] CPU AMD Ryzen 7 6800U with Radeon Graphics

[0187] Memory 64GB RAM (2100MHz)

[0188] The operating system for which the software was developed

[0189] Distributor ID: Ubuntu

[0190] Description: Ubuntu 22.04.3LTS

[0191] Release:22.04

[0192] Codename: jammy

[0193] Software development environment / development tools: Vim 8.2

[0194] Programming language: Python 3.

[0195] The output module of the system, software or device of the present invention is used to output the probability that the sample to be tested is cancer based on the prediction result of the cancer prediction module.

[0196] discuss

[0197] Tumor pathological diagnosis is one of the gold standards for diagnosing a variety of tumors at present, with a high degree of accuracy. Through tumor pathological diagnosis, the nature of the tumor can be clarified, the source of the tumor can be determined, the tumor can be staged and typed, and its degree of differentiation can be evaluated. By observing and analyzing the histological characteristics, grading and staging of the tumor, the patient's survival rate and recurrence risk can be predicted, providing an important reference for treatment decisions and patient management, thereby guiding individualized treatment strategies. In addition, tumor pathological diagnosis has a certain degree of subjectivity, and requires high professional knowledge and experience of doctors. Different pathologists may have different interpretations and judgments on the same tissue specimen. This may lead to differences in consistency of diagnosis between different hospitals and different experts, thereby affecting the comparability of treatment decisions and results. For this reason, the technicians of the present invention recognize the need to obtain more specific markers or biological characteristics of the organization from the molecular biology level to improve the accuracy of diagnosis.

[0198] In the human genome, more than 95% of genes have more than one transcript, and the same gene usually expresses different transcripts in different tissues or disease states. These different transcripts differ in structure and function. This diversity of transcripts is particularly evident in different tissues, cell types, and disease states. In particular, in the study of human cancer, the expression pattern of transcripts shows significant tissue and pathological state preferences. Specific RNA transcripts (SRT) are transcripts that are specifically expressed by genes in a certain tissue or disease. Tumor-specific RNA transcripts can not only promote the occurrence and development of cancer and become a highly anticipated target for treatment, but are also considered to be the source of new immunogenic antigens. Tissue-specific transcripts show great potential in treatment, maximizing the efficacy in target tissues and minimizing safety risks to unrelated tissues.

[0199] With the continued development of high-throughput RNA sequencing technology and the significant improvement of computational analysis methods, researchers are now able to conduct a deeper analysis of transcriptional regulatory mechanisms under normal physiological and pathological conditions, and are expected to identify new transcripts. Although short-read RNA-seq technology has been widely used in transcriptomics research, it has inherent limitations in reconstructing transcripts, especially when multiple transcript isoforms of the same gene share exons, which makes it difficult to accurately distinguish and identify. Given the high complexity of the transcriptome and the above challenges, the task of fully reconstructing all transcripts relying solely on short-read data is particularly arduous. In contrast, long-read sequencing technology (LR-Seq), also known as third-generation sequencing technology, includes Pacific Biosciences (PacBio) and Oxford Nanopore Technologies (ONT), which has the advantage of producing longer reads (>10kb) and can accurately capture full-length isoforms. Third-generation sequencing technology facilitates comprehensive and accurate characterization of transcript structure by eliminating the need for transcript assembly. Thanks to its high accuracy in capturing complete transcript isoforms, the third-generation sequencing technology provides a detailed molecular basis for studying the transcriptome characteristics of tissues and cancers. There are also many studies, including our own, that have performed long-read RNA sequencing on tissue and tumor samples and discovered unannotated tissue and tumor-specific transcripts.

[0200] Specific RNA transcripts that are expressed in tumors but not in normal tissues or are extremely low in normal tissues are called tumor SRT (tu-SRT), or tumor-specific RNA transcripts. In the complex biological process of cancer, tumor cells exhibit a series of unique transcriptome patterns. Tumor-specific RNA transcripts play a key role in regulating the biological behavior of tumors. Their presence is significantly correlated with the clinical outcomes of various cancers, and their characteristics of specific expression in tumor tissues can be used for tumor diagnosis.

[0201] Therefore, the present invention uses long-read and short-read RNA sequencing technologies to analyze tumor and normal tissue samples, and integrates RNA sequencing third-generation sequencing data from a variety of normal and cancerous human tissues and cells, identifying a large number of previously unannotated tumor-specific RNA transcripts in large-scale sample sets. These results have laid a solid scientific foundation for further research and implementation of precise RNA diagnosis and treatment strategies.

[0202] All documents mentioned in the present invention are cited as references in this application, just as each document is cited as reference individually. In addition, it should be understood that after reading the above teachings of the present invention, those skilled in the art can make various changes or modifications to the present invention, and these equivalent forms also fall within the scope defined by the claims attached to this application.

Claims

1. A method for constructing a cancer prediction model, characterized in that: Includes steps: (s1) Process the third-generation sequencing data and RNA sequencing (RNA-seq) data of cancer tissue samples, paracancerous tissue samples, and normal tissue samples to obtain reference GTF files of full-length RNA transcripts and tumor-specific RNA transcripts (Tu-SRT); (s2) providing training set, test set and validation set data for model construction, wherein the data includes expression information of full-length RNA transcripts of RNA-seq data obtained according to the reference GTF file after RNA sequencing; the expression information includes expression information of the tumor-specific RNA transcripts; Wherein, the expression information of the tumor-specific RNA transcripts includes the expression information of the tumor-specific RNA transcripts from the cancer-positive samples and the expression information of the tumor-specific RNA transcripts from the cancer-negative samples; (s3) using the training set data to train a machine learning model selected from the group consisting of a random forest model, a decision tree model, an XGBoost model, a LightGBM model, and a CatBoost model; thereby obtaining hyperparameters of the machine learning models, and obtaining a trained machine learning model; and (s4) Using the test set and validation set data, the trained machine learning model is scored for the prediction results of the machine learning model using the following scoring indicators: AUC value, accuracy, sensitivity and specificity; the machine learning model whose score meets the expected value is used as a cancer prediction model.

2. The method according to claim 1, characterized in that The expression information of the transcript includes the expression amount of the transcript measured in the form of TPM (Transcript per million).

3. The method according to claim 1, characterized in that The cancer samples include samples of n different cancers, wherein n is a positive integer ≥10; preferably, n ≥15; preferably, n is 20-100; more preferably, n is 20-50.

4. The method according to claim 1, characterized in that In step (s1), the following steps are included: (a) Collect the third-generation sequencing data and RNA-seq data of cancer tissues and adjacent tissue samples from different tumor sources, as well as normal tissue samples, and generate non-redundant transcriptome GTF files; (b) optimizing the non-redundant transcriptome GTF file, wherein, for a certain cancer / tissue RNA-seq, if all splicing sites of a transcript are detected in ≥m samples, the transcript is retained, wherein m is 1-90%, preferably 2-50%; (c) Each transcript was compared with the reference transcriptome (GENCODE v.35), and the transcripts were mainly divided into four groups, including full splicing matches, incomplete splicing matches, novel in the catalog, and novel out of the catalog. After filtering, a reference GTF file for full-length RNA transcripts was established: Among them, for all incomplete splicing matching transcripts, such transcripts are filtered out from the entire transcriptome; Filter out unqualified novel transcripts in the catalog and outside the catalog (for example, cage peaks and polyA sites were annotated using SQANTI3, and novel transcripts in the catalog and outside the catalog were filtered when there were 16 or more adenines within 20 bases downstream of the annotated transcription termination site); For similar transcripts, the longest transcript is retained and other transcripts are filtered out; (d) Based on the reference GTF file for the full-length RNA transcript, quantifying the full-length transcript of the tissue sample RNA-seq data, thereby obtaining expression information of the full-length RNA transcript of the tissue sample. (e) obtaining, by comparison, RNA transcripts having significantly different expression information between the cancer tissue sample and adjacent tissue samples and normal tissue samples, thereby obtaining the tumor-specific RNA transcripts.

5. A reference GTF file for full-length RNA transcripts, characterized in that: The steps include the following: (i) Collect the third-generation sequencing data and RNA-seq data of cancer tissues and adjacent tissue samples from different tumor sources, as well as normal tissue samples, and generate non-redundant transcriptome GTF files; (ii) optimizing the non-redundant transcriptome GTF file, wherein, for a certain cancer / tissue RNA-seq, if all splicing sites of a transcript are detected in ≥m samples, the transcript is retained, where m is 1-90%, preferably 2-50%; (iii) Each transcript was compared with the reference transcriptome (GENCODE v.35), and the transcripts were mainly divided into four groups, including full splicing matches, incomplete splicing matches, novel in the catalog, and novel out of the catalog. After filtering, a reference GTF file for full-length RNA transcripts was established: Among them, for all incomplete splicing matching transcripts, such transcripts are filtered out from the entire transcriptome; Filter out unqualified novel transcripts in the catalog and outside the catalog (for example, cage peaks and polyA sites were annotated using SQANTI3, and novel transcripts in the catalog and outside the catalog were filtered when there were 16 or more adenines within 20 bases downstream of the annotated transcription termination site); For similar transcripts, the longest transcript is retained and other transcripts are filtered out.

6. A method for determining the expression information of a full-length RNA transcript, characterized in that: Includes steps: (i) Based on the reference GTF file for full-length RNA transcripts as described in claim 5, full-length transcript quantification is performed on tissue sample RNA-seq data to obtain full-length RNA transcript expression information.

7. A cancer prediction system, characterized in that: The system comprises: (a) an input module, wherein the input module is configured to input the full-length RNA transcript expression information of the sample to be tested; (b) an evaluation module: the evaluation module is configured to input the full-length RNA transcript expression information into the cancer prediction model constructed using the method of claim 1, thereby obtaining a cancer prediction result of the sample to be tested; and (c) An output module, wherein the output module is configured to output the prediction result.

8. The system according to claim 7, characterized in that The system also includes (a0) a preprocessing module, which is configured to analyze and determine the full-length RNA transcript expression information of the sample to be tested based on the RNA-seq sequencing information of the sample to be tested.

9. The system according to claim 7, characterized in that The judgment in the prediction module includes: The full-length RNA transcript expression information is input into the cancer prediction model to obtain a prediction result, which includes a prediction probability; when the prediction probability in the prediction result is ≥ a preset reference threshold for cancer prediction, it is judged to be positive, that is, the sample to be tested is a cancer sample; otherwise, it is judged to be negative, that is, the sample to be tested is a non-cancer sample.

10. A computer storage medium, characterized in that: The storage medium is used to store a computer program corresponding to the algorithm of the cancer prediction model constructed by the method according to claim 1.

Images