Copy number variation detection method based on high-throughput transcriptome sequencing data of single sample

Abstract

A copy number variation detection method based on high-throughput transcriptome sequencing data of a single sample. The detection method comprises: first, aligning genome sequencing data to a human reference genome, and calculating the amount of data aligned to reads on each gene, so as to obtain an expression matrix; and then inputting the expression matrix into a detection model, so as to obtain a copy number variation detection result, wherein the detection model is obtained by means of performing training on the basis of a deep neural network model and by using pre-processed known database samples as a data set. By means of the detection method, the genome structure and variation of a single individual are better understood; with lower costs and in less time, gene expression level information is provided, and copy number information of genes can also be acquired, such that functions and regulatory mechanisms of genomes and the impact of copy number variations on gene expression can be better understood; and the method is of great significance for research on genetic diseases, rare diseases and complex diseases.

Description

A copy number variation detection method based on single-sample high-throughput transcriptome sequencing data Technical Field

[0001] The present invention relates to the field of bioinformatics analysis technology, and in particular to a copy number variation detection method based on single-sample high-throughput transcriptome sequencing data. Background Art

[0002] Copy Number Variations (CNVs) refer to changes in the number of copies of a region on a chromosome, meaning an increase or decrease in the number of DNA repeats in that region. CNVs are common in the human genome and have been linked to the development and progression of several human diseases.

[0003] CNVs can involve large genomic segments, even entire genes or multiple genes, thereby affecting gene expression and function. They may have a significant impact on human disease susceptibility, risk, and clinical manifestations. Some CNVs are closely associated with the development of genetic diseases, such as certain hereditary cancers, neurodevelopmental disorders (such as autism and intellectual disability), and certain congenital heart diseases.

[0004] Furthermore, CNVs can also affect drug responses and individual responsiveness to drug treatment. Some CNVs can lead to changes in the quantity or function of drug-metabolizing enzymes or drug targets, thereby affecting drug metabolism and efficacy in the body.

[0005] Therefore, detecting disease-associated CNVs and further understanding the contribution and mechanisms of these variants to disease can help improve our understanding of diseases, develop personalized medicine, and improve disease prevention and treatment methods.

[0006] There are three main strategies for CNV analysis, namely whole genome sequencing (WGS), whole exome sequencing (WES), and targeted sequencing, and detection involves a variety of algorithms and methods. The following are some common algorithms:

[0007] 1. Read depth analysis: This method infers copy number variation based on the density of sequencing reads across the genome. By comparing the read depth of the sample with that of the reference genome, regions of copy number gain or loss can be identified. However, read depth analysis is limited in detecting smaller CNVs and is susceptible to factors such as sequencing depth and regional GC content.

[0008] 2. Breakpoint analysis (split reads): This method detects CNVs by analyzing the breakpoint locations of copy number variations. It can use paired-end sequencing data or long-read sequencing data to identify breakpoint regions and infer the location and magnitude of copy number variations. However, breakpoint analysis requires high-quality sequencing data and precise breakpoint location, and is challenging for complex structural variants.

[0009] 3. Segmentation-based methods: These methods divide the genome into contiguous segments and compare the read depth or other characteristics of each segment. By detecting copy number variations between segments, the presence of CNVs can be determined. However, segmentation-based methods can have issues with false positives or false negatives when identifying small CNVs and complex structural variants.

[0010] In contrast, there are few detection solutions for CNV in transcriptome data, possibly for the following reasons:

[0011] 1) Transcriptome sequencing data inherently contain noise and technical errors, such as sequencing errors, alignment errors, and expression estimation errors. These errors may affect the results of CNV detection and require appropriate data correction and correction. Traditional CNV analysis methods may not be applicable.

[0012] 2) Gene expression signals and CNV signals in the data may interfere with each other, causing the detection of copy number variations to be interfered with and affected by gene expression;

[0013] 3) RNA-seq provides relatively sparse coverage of the genome, meaning some regions may have high coverage while others may be under-covered. This uneven coverage can affect the accuracy and sensitivity of CNV detection, making it extremely challenging, or even impossible, to detect precise CNV breakpoints and small CNV fragments using comparative depth-based methods.

[0014] 4) Current sequencing analysis of CNVs typically relies on the construction of a reference baseline. A reference baseline is a large number of individuals sequenced and analyzed to determine genomic variation in the normal population. By comparing with the reference baseline, variations in an individual's genome can be determined and the presence and copy number of CNVs inferred. However, the construction of a reference baseline also has some limitations, such as:

[0015] i) Sample size and diversity limitations: The quality and representativeness of the reference baseline depends on the number and diversity of samples included. If the reference baseline sample size is limited or lacks sufficient diversity, some population-specific CNVs may not be accurately captured.

[0016] ii) Detection of rare and individual-specific CNVs: Reference baselines typically focus on common CNV variants. However, variations between samples and datasets can exist and may not provide accurate baseline information for rare or individual-specific CNVs. These variants may have a significant impact on an individual's disease susceptibility and phenotypic characteristics.

[0017] iii) Changes in the experimental environment: Changes in the experimental environment may include changes in temperature, humidity, light intensity, and other factors, or changes in experimental equipment or reagent batches. These changes may lead to inconsistent experimental conditions, resulting in significant differences between the samples used to construct the baseline and the samples actually analyzed, which can significantly affect the analytical results.

[0018] Therefore, current transcriptome sequencing data is mostly used to detect the expression of genes and transcripts to estimate gene activity, or to identify single nucleotide polymorphisms (SNPs) and short insertions and deletions. However, it contains a large amount of information about genomic variations in samples and is not fully utilized. Among these variations, copy number variations (CNVs) are very important for cancer research because they are the main genetic drivers of cancer. However, identifying CNVs from RNA-seq data is very challenging because the dynamic and highly uneven coverage of the genome by RNA-seq signals makes it difficult to distinguish between deletion and amplification events and dynamic changes in gene expression levels.

[0019] Therefore, traditional CNV analysis methods that rely solely on reference baselines and depth-based methods may have certain limitations when used on transcriptome data, and a flexible and accurate method for detecting CNVs is urgently needed.

[0020] Summary of the Invention

[0021] The purpose of the present invention is to provide a copy number variation detection method based on single-sample high-throughput transcriptome sequencing data, which overcomes the limitations of traditional CNV analysis methods and achieves flexible and accurate CNV detection.

[0022] In view of this, the scheme of the present invention is as follows:

[0023] A copy number variation detection method based on single-sample high-throughput transcriptome sequencing data comprises the following steps:

[0024] Align the genome sequencing data to the human reference genome, calculate the amount of sequenced fragments aligned to each gene, and obtain the expression matrix;

[0025] The expression matrix is ​​input into the detection model to obtain the copy number variation detection results;

[0026] The detection model is based on a deep neural network model and is trained using preprocessed known database samples as a data set; the preprocessing includes standardizing the gene expression levels of the database samples and classifying the copy number variation types.

[0027] Furthermore, the genome sequencing data is preprocessed before the data comparison to remove low-quality sequences and cut off continuous low-quality bases.

[0028] Furthermore, before training, the database samples convert the copy number type into a numerical value that can be used by the deep learning algorithm.

[0029] Furthermore, the deep network model initializes the weights and biases of the neural network using random initialization.

[0030] Furthermore, during the training of the detection model, the hidden layer is activated to map the sample gene expression level to max(0,x), that is, when x is greater than 0, x is output, otherwise 0 is output.

[0031] Furthermore, during the training of the detection model, the output layer is activated, the input vector is normalized, and the probability value of each element is taken.

[0032] Furthermore, a cross entropy loss function is used in the detection model training process and is minimized.

[0033] The present invention also provides a copy number variation detection system based on single-sample high-throughput transcriptome sequencing data, comprising:

[0034] Computational alignment module: Aligns genome sequencing data to the human reference genome, calculates the amount of sequenced fragments aligned to each gene, and obtains an expression matrix;

[0035] Detection module: Input the expression matrix into the detection model to obtain the copy number variation detection results;

[0036] The detection model is based on a deep neural network model and is trained using preprocessed known database samples as a data set; the preprocessing includes standardizing the gene expression levels of the database samples and classifying the copy number variation types.

[0037] The present invention also provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above-mentioned detection method when executing the computer program.

[0038] The present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program implements the steps of the above-mentioned detection method when executed by a processor.

[0039] Compared with the prior art, the beneficial effects of the present invention include but are not limited to:

[0040] The detection method proposed in this paper uses a single sample for CNV analysis, enabling a better understanding of the genomic structure and variation of a single individual. It provides information not only on gene expression levels but also on gene copy number at a lower cost and time, enabling a better understanding of genomic function and regulatory mechanisms, as well as the impact of copy number variation on gene expression. The detection method can identify potential disease-associated CNVs, thereby aiding in disease diagnosis and prognosis. This has significant implications for the research of genetic diseases, rare diseases, and complex disorders. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] FIG1 is a flow chart of the method for detecting copy number variation in single-sample high-throughput transcriptome sequencing data according to the present invention. DETAILED DESCRIPTION

[0042] In order to make the purpose, technical solution and beneficial technical effects of the present invention more clear, the present invention is further described in detail below in conjunction with specific embodiments. It should be understood that the specific embodiments described in this specification are only for the purpose of explaining the present invention and are not intended to limit the present invention.

[0043] Using a single sample for CNV analysis can address several pain points and limitations, including:

[0044] Difficulty in sample collection: Traditional CNV analysis often requires a large number of samples to produce meaningful results. However, in some cases, obtaining a sufficient number of samples can be difficult or expensive. Using single samples for CNV analysis can overcome this sample collection issue, making analysis more convenient and feasible.

[0045] High cost and time to establish a reference baseline: Traditional CNV analysis often requires long sample collection times and high analysis costs. Using a single sample for analysis can save the time and cost of sample collection and obtain results faster.

[0046] To address the cost issues associated with sequencing a large number of samples and the need to construct a baseline for copy number variation detection using transcriptome data, as well as the inability of traditional DNA CNV detection methods to be applied to RNAseq data, a method for copy number variation detection based on high-throughput genome capture sequencing data from a single sample is proposed in one embodiment. The flowchart is shown in FIG1 , and specifically includes the following steps:

[0047] 1. The patient's genome sequencing data was processed using fastp software to improve data quality, remove low-quality sequences, excise continuous low-quality bases, and then align the high-quality sequences to the human reference genome hg19.

[0048] 2. Next, use a gene annotation file (such as a GTF or GFF file) to determine the location of each gene. The gene annotation file provides information about the location of each gene's transcripts and exons. Based on the alignment results, reads are associated with genes and the number of reads mapped to each gene is calculated to generate an expression matrix.

[0049] 3. We then downloaded sample data from a public database. Each sample contained gene expression levels and corresponding copy number variation types, which we used to build the model. The specific steps were:

[0050] 1) Data preprocessing: Calculate the mean (μ) and standard deviation (σ) of each gene expression (x). For each sample gene, standardize using the following formula: z = (x-μ) / σ;

[0051] 2) Categorize the copy number types into normal (copy number equal to 2), deletion (copy number less than 2), and gain (copy number greater than 2), and use one-hot encoding to convert the copy number types into numerical representations that can be used by deep learning algorithms.

[0052] 4. Then the gene expression and copy number data obtained by the above processing are divided into a training set and a test set.

[0053] 1) Initialize network parameters: Initialize the weights and biases of the neural network using random initialization.

[0054] 2) Set up an input layer (with 3 features), two hidden layers (each with 16 neurons), and an output layer (with 3 neurons)

[0055] 3) Activate the hidden layer: Use the ReLU activation function to activate the hidden layer and map the input x to max(0,x). That is, when x is greater than 0, output x, otherwise output 0.

[0056] 4) Activate the output layer: Use the Softmax function to activate the output layer. Normalize the input vector and convert the value of each element into a probability value between 0 and 1, and the sum of all elements is 1.

[0057] 5) Parameter optimization: The cross entropy loss function used in the model training process is optimized and minimized using the Adam optimizer to improve the accuracy of the model.

[0058] 6) Verify the model: Use the test set as the input of the model to verify the accuracy of the model, as shown in Table 1.

[0059] Table 1:

[0060] 5. The last step is to input the expression moment of the sample into the above model to obtain the copy number prediction results of each gene.

[0061] In the above examples, the detection method can be used to detect the presence or absence of copy number variation in a single chromosome region in a single sample and is not intended to directly diagnose specific diseases or multiple diseases. However, as a single-sample CNV result, it can help better understand the function and regulatory mechanisms of the genome, as well as the impact of copy number variation on gene expression, which is of great significance for the study of genetic diseases, rare diseases, and complex diseases.

[0062] The following is an example of CNV detection using genome sequencing data as an example:

[0063] (1) Sequencing data preprocessing

[0064] The fastq data were obtained and the statistics are shown in Table 2.

[0065] Table 2:

[0066] (2) Fastq data processing

[0067] After quality control, high-quality sequences were obtained, and the data statistics are shown in Table 3.

[0068] Table 3:

[0069] (3) Sequence alignment with reference genome

[0070] The alignment of the sequence data with the human reference genome hg19 is shown in Table 4.

[0071] Table 4:

[0072] (4) Calculate the number of reads for each gene, as shown in Table 5.

[0073] Table 5:

[0074] (5) Input the detection model to obtain the copy number variation detection results, as shown in Table 6.

[0075] Table 6:

[0076] The present invention is not limited to what is described in the specification and embodiments, and additional advantages and modifications will be readily apparent to those skilled in the art. Therefore, the present invention is not limited to the specific details, representative schemes, and examples described herein without departing from the spirit and scope of the general concept defined by the claims and their equivalents.

Claims

1. A copy number variation detection method based on single sample high-throughput transcriptome sequencing data, characterized in that: The steps include: Align the genome sequencing data to the human reference genome, calculate the amount of data aligned to the sequencing fragments on each gene, and obtain the expression matrix; The expression matrix is ​​input into the detection model to obtain the copy number variation detection results; The detection model is based on a deep neural network model and is trained using preprocessed known database samples as a data set; the preprocessing includes standardizing the gene expression levels of the database samples and dividing the copy number variation types.

2. The detection method according to claim 1, characterized in that: Before the data comparison, the genome sequencing data is preprocessed to remove low-quality sequences and cut off continuous low-quality bases.

3. The detection method according to claim 1, characterized in that: Before training, the database samples convert the copy number type into a numerical value that can be used by the deep learning algorithm.

4. The detection method according to claim 1, characterized in that: The deep network model initializes the weights and biases of the neural network in a random initialization manner.

5. The detection method according to claim 1, characterized in that: During the training of the detection model, the hidden layer is activated to map the sample gene expression to max(0,x), that is, when x is greater than 0, x is output, otherwise 0 is output.

6. The detection method according to claim 1, characterized in that: During the training process of the detection model, the output layer is activated, the input vector is normalized, and the probability value of each element is taken.

7. The detection method according to claim 1, characterized in that: The cross entropy loss function is used in the detection model training process and is minimized.

8. A copy number variation detection system based on single sample high-throughput transcriptome sequencing data, characterized in that: include: Computational alignment module: Align the genome sequencing data to the human reference genome, calculate the amount of data aligned to the sequencing fragments on each gene, and obtain the expression matrix; Detection module: input the expression matrix into the detection model to obtain the copy number variation detection results; The detection model is based on a deep neural network model and is trained using preprocessed known database samples as a data set; the preprocessing includes standardizing the gene expression levels of the database samples and dividing the copy number variation types.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images