Method and device for detecting abnormity of copy number of prenatal fetus

By performing multi-level correction and sliding window judgment on genome sequencing data, the problem of low accuracy in fetal chromosomal disease detection in the prior art is solved, and high accuracy detection of fetal copy number abnormalities is achieved, which reduces the harm of pregnant women and False positive rate.

CN118983010APending Publication Date: 2024-11-19BEIJING USCI MEDICAL DEVICES CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411044280.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-31
Publication Date
2024-11-19

AI Technical Summary

Technical Problem

In the prior art, the prenatal detection accuracy of fetal chromosomal diseases is not high, the serological screening accuracy is low, and the false positive rate is high. Although amniotic fluid puncture is accurate, it is very traumatic to pregnant women, and there is a risk of miscarriage, making it difficult to achieve large-scale Prenatal testing.

Method used

By dividing the genome sequencing data into windows of the same length, basic correction, data quantity correction and PCA correction are performed, parent CNV is removed, and copy number variation type and variation region are judged using Z-value calculation and sliding window method to improve the accuracy of detection.

Benefits of technology

It realizes accurate detection of fetal copy number abnormalities, reduces false positive rates and injury to pregnant women, and is suitable for large-scale prenatal testing.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118983010A_ABST
    Figure CN118983010A_ABST
Patent Text Reader

Abstract

The invention provides a method and a device for detecting copy number abnormity of a prenatal fetus. The method comprises the following steps: S1, dividing genome sequencing data into windows with the same length, and obtaining the original reading number of each window in a first to-be-detected sample and a first negative sample; s2, sequentially performing basic correction, data volume correction and PCA correction on the original reading number of each window in the first to-be-detected sample and the first negative sample; s3, removing a parent CNV in the sequence number of each window in the first to-be-detected sample after PCA correction, and performing data volume correction again; and S4, calculating a Z value by using the sequence number of each window in the first negative sample after PCA correction and the sequence number of each window in the first to-be-detected sample after correction, and judging the copy number variation type and variation region of the to-be-detected sample by using the Z value through a sliding window method, and an accurate detection result of the copy number variation type of the to-be-detected sample can be obtained.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of gene detection technology, and in particular to a method and device for detecting abnormal copy number of a prenatal fetus. Background Art

[0002] Non-invasive prenatal screening (NIPS) uses molecular genetic technologies such as high-throughput sequencing to detect fetal free DNA fragments in maternal plasma during pregnancy to assess the risk of common chromosomal aneuploidy and other genomic diseases in the fetus.

[0003] Studies have shown that trisomy 21, 18, and 13 has an incidence rate of 0.2%, and the proportion of fetuses carrying clinically significant pCNVs in pregnant women can reach 1.6%-1.7%. It is one of the important genetic causes of birth defects such as congenital malformations and intellectual disabilities. Therefore, prenatal screening and prenatal diagnosis of the fetus are of great significance for the prevention and control of birth defects.

[0004] Prenatal testing for chromosomal diseases mainly consists of serological screening and amniocentesis. However, serological screening has low accuracy, with a 5% false positive rate and a 20-40% missed diagnosis rate. Although amniocentesis has high accuracy, it is traumatic to pregnant women and has a 1% risk of miscarriage, which is not convenient for large-scale prenatal testing. Summary of the invention

[0005] The main purpose of the present invention is to provide a method and device for detecting abnormal fetal copy number before birth, so as to solve the problem of low accuracy in detecting fetal chromosomal diseases in the prior art.

[0006] In order to achieve the above-mentioned object, according to a first aspect of the present invention, a method for detecting abnormal copy number of a prenatal fetus is provided, the method comprising: S1, dividing genome sequencing data into windows of the same length, counting the number of reads of sequencing data of the test sample and the negative sample in each window, and obtaining the number of original reads in each window of the first test sample and the first negative sample; S2, sequentially performing basic correction, data volume correction and PCA correction on the number of original reads in each window of the first test sample and the first negative sample, and obtaining the number of sequences in each window of the first test sample and the first negative sample after PCA correction; S3, removing the maternal CNV from the number of sequences in each window of the first test sample after PCA correction, and re-performing data volume correction to obtain the number of sequences in each window of the first test sample after correction; S4, calculating the Z value using the number of sequences in each window of the first negative sample after PCA correction and the number of sequences in each window of the first test sample after correction, and using the Z value to determine the copy number variation type and variation region of the test sample through the sliding window method; wherein the basic correction includes correction of GC content and comparability.

[0007] Furthermore, the Z value is calculated using the following formula I: Zscore = hh test -mean(hh) / sd(hh), formula Ⅰ; where hh test is the average value of the content of all windows covered by the area to be detected in the first sample to be tested after correction; mean(hh) is the average value of the number of sequences of all windows covered by the area to be detected in the first negative sample after PCA correction; sd(hh) is the variance of the number of sequences of all windows covered by the area to be detected in the first negative sample after PCA correction.

[0008] Further, S4 includes: S41, constructing a table of behavioral Z values ​​and columns as copy number states, where the copy number states include "copy number loss" and "copy number duplication"; S42, using the calculated Z value to determine the copy number state of the initial window to be detected, and filling the Z value and copy number state of the initial window to be detected into the table; S43, starting from the end coordinates of the initial window to be detected in the sample to be detected and ending at the end coordinates of the last window, sliding backward in units of the length of a window, performing a sliding window judgment each time sliding to one unit, and obtaining the copy number variation type and variation region of the sample to be detected; wherein the sliding window judgment includes: using the calculated Z value to determine the copy number state of the window to be determined, and filling the Z value and copy number state of the window to be determined into the table; the initial window to be detected is a window with a length of one unit after the end coordinates of the first window, each normal window, copy number loss region or copy number duplication region in the sample to be detected.

[0009] Further, S42 includes: if the Z value is ≤-3, the initial window to be detected is a copy number deletion window, the starting coordinates of the window to be detected are determined as the starting coordinates of the copy number deletion region, and the Z value is filled in the "Copy Number Deletion" column of the table; if the Z value is ≥3, the initial window to be detected is a copy number duplication window, the starting coordinates of the window to be detected are determined as the starting coordinates of the copy number duplication region, and the Z value is filled in the "Copy Number Duplication" column of the table; if the Z value is less than 3 and greater than -3, the initial window to be detected is a normal window and does not carry copy number variation.

[0010] Further, S43 includes: if the initial window to be detected is a normal window, the start and end coordinates of the area to be detected are both moved backward by one unit length, and the operation of S42 is repeated until the copy number variation is detected or all windows are detected; if the initial window to be detected is a copy number deletion window, the start coordinates of the copy number deletion area remain unchanged, the end coordinates move backward by one unit length, and the sliding window method judgment operation is repeated for the subsequent single window to be judged until the Z value of the window to be detected is greater than the minimum value +1 of the "copy number deletion" column in the table, then the end coordinates of the window to be detected are the end coordinates of the copy number deletion area; if the initial window to be detected is a copy number duplication window, the start coordinates of the copy number duplication area remain unchanged, the end coordinates move backward by one unit length, and the sliding window method judgment operation is repeated for the subsequent single window to be judged until the Z value of the window to be detected is greater than the maximum value -1 of the "copy number duplication" column in the table, then the end coordinates of the window to be detected are the end coordinates of the copy number deletion area.

[0011] Furthermore, the test sample and the negative sample include: peripheral blood of pregnant women; preferably, the length of the window is 50-100kb; preferably, the basic correction includes: using the local weighted regression method to perform GC content correction and comparability correction on the first test sample and the first negative sample respectively, to obtain the number of sequences in each window after basic correction in the first test sample and the first negative sample; preferably, the proportion of the number of sequences in each window after basic correction to the number of sequences in all windows after basic correction is calculated, and the total data volume is 40-50M, and the number of sequences in each window after basic correction is proportionally calculated in the total data volume to obtain the number of sequences in each window after data volume correction; preferably, the maternal CNV in the number of sequences in each window in the first test sample after PCA correction is removed using a hidden Markov chain.

[0012] Furthermore, the PCA correction includes: using the number of sequences of each window of the first negative samples greater than 10 after data amount correction to construct a data reference set, performing PCA training on the reference set to form a PCA reference set; removing the first 10 principal components in the PCA reference set from the number of sequences of each window of the first negative sample and the first sample to be detected after data amount correction, and obtaining the number of sequences of each window in the first negative sample and the sample to be detected after PCA correction, respectively.

[0013] In order to achieve the above-mentioned purpose, according to a second aspect of the present invention, there is provided a device for detecting abnormalities in prenatal fetal copy number, the device comprising: a window segmentation module, configured to divide genome sequencing data into windows of the same length, count the number of reads of sequencing data of the test sample and the negative sample in each window, and obtain the original number of reads in each window of the first test sample and the first negative sample; a correction module, configured to perform basic correction, data amount correction and PCA correction on the original number of reads in each window of the first test sample and the first negative sample in turn, and obtain the original number of reads in each window of the first test sample and the first negative sample after PCA respectively. The number of sequences after correction; the maternal CNV removal module is configured to remove the maternal CNV from the number of sequences in each window of the first sample to be tested after PCA correction, and re-correct the data volume to obtain the corrected number of sequences in each window of the first sample to be tested; the sliding window detection module is configured to calculate the Z value using the number of sequences in each window of the first negative sample after PCA correction and the corrected number of sequences in each window of the first sample to be tested, and use the Z value to determine the copy number variation type and variation region of the sample to be tested through the sliding window method; wherein the basic correction includes correction of GC content and comparability.

[0014] Furthermore, the sliding window detection module includes a Z value calculation unit: the Z value calculation unit is configured to calculate the Z value using the following formula I; Zscore = hh test -mean(hh) / sd(hh), formula Ⅰ; where h htest is the average value of the content of all windows covered by the area to be detected in the first sample to be tested after correction; mean(hh) is the average value of the number of sequences of all windows covered by the area to be detected in the first negative sample after PCA correction; sd(hh) is the variance of the number of sequences of all windows covered by the area to be detected in the first negative sample after PCA correction.

[0015] Further, the sliding window detection module includes: a table construction unit, which is configured to construct a table with Z values ​​as rows and copy number states as columns, and the copy number states include "copy number deletion" and "copy number duplication"; a detection unit for the initial window to be detected, which is configured to use the calculated Z value to determine the copy number state of the initial window to be detected, and fill the Z value and copy number state of the window to be detected into the table; an output copy number variation unit, which is configured to start from the end coordinates of the first window in each area to be detected in the sample to be detected and end at the end coordinates of the last window, and slide backward in sequence in units of the length of a window, and perform a sliding window method judgment every time it slides to one unit to obtain the copy number variation type and variation area of ​​the sample to be detected; wherein the sliding window method judgment includes: using the calculated Z value to determine the copy number state of the window to be determined, and filling the Z value and copy number state of the window to be determined into the table; the initial window to be detected is a window with a length of one unit after the end coordinates of the first window, each normal window, copy number deletion area or copy number duplication area in the sample to be detected.

[0016] Further, the unit for detecting the initial window to be detected includes: a copy number deletion judgment unit, which is set to, if the Z value is ≤-3, the initial window to be detected is a copy number deletion window, the starting coordinates of the window to be detected are determined as the starting coordinates of the copy number deletion region, and the Z value is filled in the "Copy Number Deletion" column of the table; a copy number duplication judgment unit, which is set to, if the Z value is ≥3, the initial window to be detected is a copy number duplication window, the starting coordinates of the window to be detected are determined as the starting coordinates of the copy number duplication region, and the Z value is filled in the "Copy Number Duplication" column of the table; a normal judgment unit, which is set to, if the Z value is less than 3 and greater than -3, the initial window to be detected is a normal window and does not carry copy number variation.

[0017] Further, the output copy number variation unit includes: a normal copy number unit, which is set to, if the initial window to be detected is a normal window, the start and end coordinates of the area to be detected are both moved backward by a unit length, and the operation of detecting the initial window unit to be detected is repeatedly performed until the copy number variation is detected or all windows are detected; a copy number deletion unit, which is set to, if the initial window to be detected is a copy number deletion window, the start coordinates of the copy number deletion area remain unchanged, the end coordinates move backward by a unit length, and the operation of sliding window judgment is repeatedly performed on the subsequent single window to be judged until the Z value of the window to be detected is greater than the minimum value +1 of the "copy number deletion" column in the table, then the end coordinates of the window to be detected are the end coordinates of the copy number deletion area; a copy number duplication unit, which is set to, if the initial window to be detected is a copy number duplication window, the start coordinates of the copy number duplication area remain unchanged, the end coordinates move backward by a unit length, and the operation of sliding window judgment is repeatedly performed on the subsequent single window to be judged until the Z value of the window to be detected is greater than the maximum value -1 of the "copy number duplication" column in the table, then the end coordinates of the window to be detected are the end coordinates of the copy number deletion area.

[0018] Furthermore, the sample to be tested and the negative sample include: peripheral blood of pregnant women; preferably, the length of the window is 50-100kb; preferably, the correction module includes: a basic correction unit, which is configured to use a local weighted regression method to perform GC content correction and comparability correction on the first sample to be tested and the first negative sample respectively, and obtain the number of sequences in each window after basic correction in the first sample to be tested and the first negative sample; preferably, the correction module also includes: a data amount correction unit, which is configured to calculate the proportion of the number of sequences in each window after basic correction to the number of sequences in all windows after basic correction, with a total data volume of 40-50M, and calculate the number of sequences in each window after basic correction in proportion to the number of sequences in the total data volume, and obtain the number of sequences in each window after data amount correction; preferably, a hidden Markov chain is used to remove the maternal CNV in the number of sequences in each window in the first sample to be tested after PCA correction.

[0019] Furthermore, the correction module also includes: a PCA correction unit, which is configured to construct a data reference set using the number of sequences of each window of the first negative samples greater than 10 after data amount correction, perform PCA training on the reference set to form a PCA reference set; remove the first 10 principal components in the PCA reference set from the number of sequences of each window of the first negative sample and the first sample to be detected after data amount correction, and obtain the number of sequences of each window in the first negative sample and the sample to be detected after PCA correction, respectively.

[0020] In order to achieve the above-mentioned purpose, according to the third aspect of the present invention, a computer-readable storage medium is provided, the storage medium includes a stored program, wherein when the program is executed, the device where the storage medium is located is controlled to execute the above-mentioned method for detecting prenatal fetal copy number abnormalities.

[0021] In order to achieve the above object, according to a fourth aspect of the present invention, a processor is provided, the processor being used to run a program, wherein the program runs the above method for detecting prenatal fetal copy number abnormalities.

[0022] By applying the technical solution of the present invention, the sequencing data of the test sample and the negative sample are divided into windows, and basic correction, data amount correction and PCA correction are performed in sequence, the maternal CNV in the test sample is removed, the Z value of the sequence number of the test sample and the negative sample after the above correction is calculated, and the copy number variation type is judged, so that a more accurate detection result of the copy number variation type of the test sample can be obtained. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] The drawings constituting a part of the present application are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0024] Figure 1 A flow chart of the detection method of the present application is shown;

[0025] Figure 2 The result diagram of GC content correction in Example 1 of the present application is shown;

[0026] Figure 3 The result diagram of the comparability correction in Example 1 of the present application is shown. DETAILED DESCRIPTION

[0027] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present invention will be described in detail below in conjunction with the embodiments.

[0028] As mentioned in the background technology, the existing detection of fetal chromosomal diseases is mainly based on serological screening and amniocentesis. However, serological screening has low accuracy, and although amniocentesis has high accuracy, it is traumatic to pregnant women and has a 1% risk of miscarriage, which is not convenient for large-scale prenatal testing. Therefore, this application intends to protect a method for detecting fetal copy number abnormalities with high accuracy and less harm to pregnant women.

[0029] In a first typical embodiment of the present application, a method for detecting prenatal fetal copy number abnormality is provided, the method comprising: S1, dividing genome sequencing data into windows of the same length, counting the number of reads in each window of the sequencing data of the test sample and the negative sample, and obtaining the original number of reads in each window of the first test sample and the first negative sample; S2, performing basic correction, data volume correction and PCA correction on the original number of reads in each window of the first test sample and the first negative sample in turn, and obtaining the number of sequences in each window of the first test sample and the first negative sample after PCA correction respectively; S3, removing the maternal CNV from the number of sequences in each window of the first test sample after PCA correction, and re-performing data volume correction to obtain the corrected number of sequences in each window of the first test sample; S4, calculating the Z value using the number of sequences in each window of the first negative sample after PCA correction and the corrected number of sequences in each window of the first test sample, and using the Z value to determine the copy number variation type and variation region of the test sample through the sliding window method; wherein the basic correction includes correction of GC content and comparability.

[0030] This application corrects the GC content, comparability, data volume, and sequencing depth in the sequencing data by window segmentation of the sequencing data of the sample to be tested and the negative sample, and further corrects the accuracy of the sequencing data. Afterwards, the maternal CNV (Copy number variation) in the sequencing data of the sample to be tested after the above correction is further removed to avoid the copy number variation of the sample to be tested from mixing into the sequencing data from the mother. And then use the sequence number of each window of the sample to be tested and the negative sample after the above correction to judge the copy number variation type of the sample to be tested. The copy number variation type and variation region obtained can further confirm whether the fetus of the sample to be tested suffers from a chromosomal disease.

[0031] In a preferred embodiment, the Z value is calculated using the following formula I:

[0032] Zscore=hh test -mean(hh) / sd(hh), formula I;

[0033] Among them, hh test is the average value of the content of all windows covered by the area to be detected in the first sample to be tested after correction; mean(hh) is the average value of the number of sequences of all windows covered by the area to be detected in the first negative sample after PCA correction; sd(hh) is the variance of the number of sequences of all windows covered by the area to be detected in the first negative sample after PCA correction.

[0034] In order to further accurately determine the type of copy number variation, the Z value of each window is determined to obtain a comprehensive test result of the sample to be tested. In a preferred embodiment, S4 includes: S41, constructing a table with Z values ​​as rows and copy number states as columns, and the copy number states include "copy number deletion" and "copy number duplication"; S42, using the calculated Z value to determine the copy number state of the initial window to be detected, and filling the Z value and copy number state of the initial window to be detected into the table; S43, starting from the end coordinates of the initial window to be detected in the sample to be detected and ending at the end coordinates of the last window, sliding backward in sequence in units of the length of a window, performing a sliding window method judgment every time a unit is slid to, and obtaining the copy number variation type and variation region of the sample to be tested; wherein the sliding window method judgment includes: using the calculated Z value to determine the copy number state of the window to be determined, and filling the Z value and copy number state of the window to be determined into the table; the initial window to be detected is a window with a length of one unit after the end coordinates of the first window, each normal window, copy number deletion region or copy number duplication region in the sample to be detected.

[0035] In order to further accurately use the Z value to determine the type of copy number variation, in a preferred embodiment, S42 includes: if the Z value is ≤-3, the initial window to be detected is a copy number deletion window, the starting coordinates of the window to be detected are determined as the starting coordinates of the copy number deletion region, and the Z value is filled in the "Copy Number Deletion" column of the table; if the Z value is ≥3, the initial window to be detected is a copy number duplication window, the starting coordinates of the window to be detected are determined as the starting coordinates of the copy number duplication region, and the Z value is filled in the "Copy Number Duplication" column of the table; if the Z value is less than 3 and greater than -3, the initial window to be detected is a normal window and does not carry a copy number variation.

[0036] In order to further accurately determine the type of copy number variation, the Z value of each window is compared to determine the copy number variation type of the sample to be tested. In a preferred embodiment, S43 includes: if the initial window to be detected is a normal window, the start and end coordinates of the region to be detected are both moved backward by one unit length, and the operation of S42 is repeatedly performed until the copy number variation is detected or all windows are detected; if the initial window to be detected is a copy number deletion window, the start coordinates of the copy number deletion region remain unchanged, the end coordinates move backward by one unit length, and the sliding window method judgment operation is repeatedly performed on the subsequent single window to be determined until the Z value of the window to be detected is greater than the minimum value +1 of the "copy number deletion" column in the table, then the end coordinates of the window to be detected are the end coordinates of the copy number deletion region; if the initial window to be detected is a copy number duplication window, the start coordinates of the copy number duplication region remain unchanged, the end coordinates move backward by one unit length, and the sliding window method judgment operation is repeatedly performed on the subsequent single window to be determined until the Z value of the window to be detected is greater than the maximum value -1 of the "copy number duplication" column in the table, then the end coordinates of the window to be detected are the end coordinates of the copy number deletion region.

[0037] Any sample type that can obtain sample sequencing data is applicable to this application. In a preferred embodiment, the sample to be tested and the negative sample include: peripheral blood of pregnant women. Any window length that can perform a relatively accurate Z value judgment is applicable to this application. In a preferred embodiment, the window length is 50-100kb.

[0038] In order to obtain more accurate sequencing data, in a preferred embodiment, before basic correction, the test samples and negative samples need to be quality controlled, and the adapter sequences, N-containing bases and bases with sequencing quality less than 15 in the base sequence of the original sequencing data are removed to generate a new fastq file; in a preferred embodiment, the quality control tool is Trimmomatic-0.39 software.

[0039] In order to obtain more accurate sequencing data, in a preferred embodiment, after quality control, it is necessary to extract the uniquely aligned base sequences in the original sequencing data and further remove the repeated sequences; in a preferred embodiment, the base sequence alignment is performed with the human reference genome sequence (version hg19); in a preferred embodiment, the software for unique alignment is BWA v0.7.17; the software for repeated sequence removal is Picard v2.20.3.

[0040] Since there is a certain GC preference in the sequencing process and the comparability of different regions of the chromosome also varies, in order to further obtain more accurate sequencing data, in a preferred embodiment, the basic correction includes: using the local weighted regression method to perform GC content correction and comparability correction on the first test sample and the first negative sample respectively, to obtain the number of sequences in each window in the first test sample and the first negative sample after basic correction.

[0041] Among them, GC correction includes: fitting a trend line between the number of sequences in each window and the GC content of each window as the baseline, subtracting the number of sequences corresponding to the baseline from the original number of sequences in each window, and adding the median number of original sequences of all windows, which is the number of sequences in the window after GC correction. Comparability correction includes: fitting a trend line between the number of sequences in each window after GC correction and the value of comparability of each window as the baseline, subtracting the number of sequences corresponding to the baseline from the number of sequences in each window after GC correction, and adding the median number of sequences in all windows after GC correction, which is the number of sequences in the window after basic correction.

[0042] Since different samples have different sequencing levels, in order to keep the data volume of each window at the same level during the subsequent PCA correction, it is necessary to perform normalization correction on the number of sequences in each window in each sample after basic correction. In a preferred embodiment, the ratio of the number of sequences in each window after basic correction to the number of sequences in all windows after basic correction is calculated. With a total data volume of 40-50M, the number of sequences in each window after basic correction in the total data volume is calculated proportionally to obtain the number of sequences in each window after data volume correction.

[0043] To further correct the data, in a preferred embodiment, the PCA correction includes: using the number of sequences of each window of the first negative samples greater than 10 after data amount correction to construct a data reference set, performing PCA training on the reference set to form a PCA reference set; removing the first 10 principal components in the PCA reference set from the number of sequences of each window of the first negative sample and the first sample to be detected after data amount correction, and obtaining the number of sequences of each window in the first negative sample and the sample to be detected after PCA correction, respectively.

[0044] Specifically, PCA correction includes: forming a matrix with the number of sequences of negative samples after data volume correction, where each column of the matrix is ​​a sample and each row is a window, and performing SVD decomposition on the matrix to extract the first 10 principal components. The number of sequences of each sample in the negative samples after data volume correction can be regarded as a vector, and the first 10 principal components are removed, and the average number of sequences of each window after the samples of all negative samples are corrected by the PCA algorithm is obtained. This ratio is called the number of sequences in this window after PCA correction in this sample.

[0045] Since humans are diploid, the copy number variation is about 50% (1 / 2) level, and the fetal DNA only occupies a part of the mother's DNA, such as f, so if it is a fetal copy number variation, it is about 50%*f. If a window contains maternal CNV, the sequence number of the window changes by about 50% compared with the window without CNV. Therefore, removing the interference of maternal CNV is very important for accurately detecting the copy number variation type of pregnant women and fetuses. In a preferred embodiment, the maternal CNV in the sequence number of each window in the first sample to be tested after PCA correction is removed using a hidden Markov chain.

[0046] Specifically, the hidden Markov chain correction includes: using the hidden state of each chromosome to form a hidden Markov chain, and iterating 10 times with the EM algorithm, fitting the model to the optimal parameters, and finally calculating the posterior probability of each window carrying the maternal CNV. When the probability of the window carrying the maternal CNV is greater than 0.99, the window is removed, otherwise it is retained. The window that does not carry the maternal CNV is re-corrected for the amount of data to obtain the number of sequences in each window in the first sample to be tested after correction.

[0047] Among them, the emission function of the hidden Markov is shown in the following formula II: (from the hidden state, i.e., CNV type, to the visible state, i.e., the number of sequences in the window):

[0048]

[0049] In the above formula, k (-1: missing, 0: negative, 1: amplification) represents the three states of maternal CNV;

[0050] x m The window content corresponding to the mth window;

[0051] Z m The states corresponding to the maternal CNV in the mth window have three possibilities: -1, 0, and 1;

[0052] σ k The variance of the window content corresponding to the three states is set to 0.05;

[0053] mk Corresponding to the average value of the window content under the three states, when k is -1, 0, and 1 respectively, m k Set to -0.5, 0, 0.5;

[0054] P(Xm|Zm=k) is the posterior probability of the three situations of maternal CNV deletion, negative, and amplification in this window.

[0055] The posterior probability when the mth window state is maternal CNV deletion, negative, or amplification (i.e., k = -1, 0, 1) is as shown in the following formula III (transition from one state to the next):

[0056] Rmk=[r*P(xm-1|zm-1=k)+(1-r)*αk]*P(xm|zm=k)

[0057] Zmk=Rmk / (Rm(k=-1)+Rm(k=0)+Rm(k=1)), formula III;

[0058] Among them, r is preset to 0.1;

[0059] When k is -1, 0, and 1, αk is preset to 0.1, 0.98, and 0.1, respectively.

[0060] When m=1, P(xm-1|zm-1=k) is preset. When k is -1, 0, and 1, P(xm-1|zm-1=k) is preset to 0.1, 0.98, and 0.1 respectively.

[0061] Assuming there are n windows in total, when m increases from 1 to n, the Zmk of the three states of each window is calculated, and the variance σ when the state is -1, 0, and 1 is obtained k , average value m k , iterate into formula II, repeat 10 times, and the Zkm calculated for the last time is the posterior probability of the three situations of missing, negative, and amplified maternal CNV in the window.

[0062] In a second typical embodiment of the present application, a device for detecting prenatal fetal copy number abnormalities is provided, the device comprising: a window segmentation module, configured to divide genome sequencing data into windows of the same length, count the number of reads of the sequencing data of the test sample and the negative sample in each window, and obtain the original number of reads in each window of the first test sample and the first negative sample; a correction module, configured to perform basic correction, data volume correction and PCA correction on the original number of reads in each window of the first test sample and the first negative sample in turn, and obtain the original number of reads in each window of the first test sample and the first negative sample after PCA correction. The invention relates to a method for detecting a copy number variation of a first sample to be tested and a method for determining the copy number variation type and variation region of a first sample to be tested. The method comprises the following steps: a maternal CNV removal module, which is configured to remove the maternal CNV from the number of sequences in each window in the first sample to be tested after PCA correction, and re-correct the data volume to obtain the corrected number of sequences in each window in the first sample to be tested; a sliding window detection module, which is configured to calculate the Z value using the number of sequences in each window in the first negative sample after PCA correction and the corrected number of sequences in each window in the first sample to be tested, and use the Z value to determine the copy number variation type and variation region of the sample to be tested through the sliding window method; wherein the basic correction includes correction of GC content and comparability.

[0063] The present application uses the window segmentation module, correction module, maternal CNV removal module and sliding window detection module in the device to perform window segmentation on the sequencing data of the sample to be tested and the negative sample, and then GC content, comparability and data volume in the sequencing data are corrected, and the accuracy of the sequencing data is further corrected. Afterwards, the maternal CNV (Copy number variation) in the sequencing data of the sample to be tested after the above correction is further removed to ensure the accuracy of the copy number variation detection of the sample to be tested. Then, the original amount of the sequence number of each window of the sample to be tested and the negative sample after the above correction is used to judge the copy number variation type of the sample to be tested. The copy number variation type and region obtained can further confirm whether the fetus of the sample to be tested suffers from a chromosomal disease.

[0064] In a preferred embodiment, the sliding window detection module includes a Z value calculation unit: the Z value calculation unit is configured to calculate the Z value using the following formula I;

[0065] Zscore=hh test -mean(hh) / sd(hh), formula I;

[0066] Among them, hh testis the average value of the content of all windows covered by the area to be detected in the first sample to be tested after correction; mean(hh) is the average value of the number of sequences of all windows covered by the area to be detected in the first negative sample after PCA correction; sd(hh) is the variance of the number of sequences of all windows covered by the area to be detected in the first negative sample after PCA correction.

[0067] In order to further accurately determine the type of copy number variation, the Z value of each window is determined to obtain a comprehensive detection result of the sample to be tested. In a preferred embodiment, the sliding window detection module includes: a table construction unit, which is configured to construct a table with rows as Z values ​​and columns as copy number states, and the copy number states include "copy number deletion" and "copy number duplication"; an initial window to be tested unit, which is configured to use the calculated Z value to determine the copy number state of the initial window to be tested, and fill the Z value and copy number state of the window to be tested into the table; an output copy number variation unit, which is configured to output the copy number variation from each window to be tested in the sample to be tested. The detection area starts from the end coordinates of the first window and ends at the end coordinates of the last window, and slides backward in units of the length of a window. Each time it slides to a unit, a sliding window method judgment is performed to obtain the copy number variation type and variation region of the sample to be tested; wherein the sliding window method judgment includes: using the calculated Z value to judge the copy number status of the window to be judged, and filling the Z value and copy number status of the window to be judged into a table; the initial window to be detected is a window with a length of one unit after the end coordinates of the first window, each normal window, copy number deletion region or copy number duplication region in the sample to be tested.

[0068] In order to further accurately use the Z value to determine the type of copy number variation, in a preferred embodiment, the unit for detecting the initial window to be detected includes: a copy number loss judgment unit, which is set to: if the Z value is ≤-3, the initial window to be detected is a copy number loss window, the starting coordinates of the window to be detected are determined as the starting coordinates of the copy number loss region, and the Z value is filled in the "copy number loss" column of the table; a copy number duplication judgment unit, which is set to: if the Z value is ≥3, the initial window to be detected is a copy number duplication window, the starting coordinates of the window to be detected are determined as the starting coordinates of the copy number duplication region, and the Z value is filled in the "copy number duplication" column of the table; a normal judgment unit, which is set to: if the Z value is less than 3 and greater than -3, the initial window to be detected is a normal window and does not carry a copy number variation.

[0069] In order to further accurately determine the type of copy number variation, the Z value of each window is compared to determine the copy number variation type of the sample to be tested. In a preferred embodiment, the output copy number variation unit includes: a normal copy number unit, which is set to, if the initial window to be detected is a normal window, the start and end coordinates of the area to be detected are both moved backward by one unit of length, and the operation of detecting the initial window unit to be detected is repeated until the copy number variation is detected or all windows are detected; a copy number deletion unit, which is set to, if the initial window to be detected is a copy number deletion window, the start coordinates of the copy number deletion area remain unchanged, and the end coordinates move backward by one unit of length The sliding window method is used to repeatedly perform the judgment operation on the subsequent single window to be judged until the Z value of the window to be detected is greater than the minimum value + 1 of the "copy number deletion" column in the table, and the end coordinates of the window to be detected are the end coordinates of the copy number deletion region; the copy number repeat unit is set to that if the initial window to be detected is a copy number repeat window, the start coordinates of the copy number repeat region remain unchanged, and the end coordinates move backward by one unit length, and the sliding window method is repeatedly performed on the subsequent single window to be judged until the Z value of the window to be detected is greater than the maximum value -1 of the "copy number repeat" column in the table, and the end coordinates of the window to be detected are the end coordinates of the copy number deletion region.

[0070] Any sample type that can obtain sample sequencing data is applicable to this application. In a preferred embodiment, the sample to be tested and the negative sample include: peripheral blood of pregnant women. Any window length that can perform a relatively accurate Z value judgment is applicable to this application. In a preferred embodiment, the window length is 50-100kb.

[0071] In order to obtain more accurate sequencing data, in a preferred embodiment, the correction module includes: a quality control unit, which is configured to perform quality control on the test sample and the negative sample before basic correction, and remove the adapter sequence, N-containing bases, and bases with a sequencing quality less than 15 in the base sequence of the original sequencing data to generate a new fastq file; in a preferred embodiment, the quality control tool is Trimmomatic-0.39 software.

[0072] In order to obtain more accurate sequencing data, in a preferred embodiment, the correction module includes: a unique alignment unit, which is configured to extract unique alignment base sequences in the original sequencing data after quality control, and further remove duplicate sequences; in a preferred embodiment, the base sequence alignment is performed with the human reference genome sequence (version hg19); in a preferred embodiment, the software for unique alignment is BWA v0.7.17; the software for duplicate sequence removal is Picard v2.20.3.

[0073] Since there is a certain GC preference in the sequencing process and the comparability of different regions of the chromosome also varies, in order to further obtain more accurate sequencing data, in a preferred embodiment, the basic correction unit is configured to use a local weighted regression method to perform GC content correction and comparability correction on the first test sample and the first negative sample respectively, to obtain the number of sequences in each window in the first test sample and the first negative sample after basic correction.

[0074] Since different samples have different sequencing levels, in order to keep the data volume of each window at the same level during the subsequent PCA correction, it is necessary to perform normalization correction on the number of sequences in each window in each sample after basic correction. In a preferred embodiment, the correction module also includes: a data volume correction unit, which is configured to calculate the ratio of the number of sequences in each window after basic correction to the number of sequences in all windows after basic correction, and calculate the number of sequences in each window after basic correction in proportion to the total data volume with a total data volume of 40-50M, so as to obtain the number of sequences in each window after data volume correction.

[0075] To further perform correction, in a preferred embodiment, the correction module also includes: a PCA correction unit, which is configured to construct a data reference set using the number of sequences of each window of the first negative samples greater than 10 after data amount correction, perform PCA training on the reference set to form a PCA reference set; remove the first 10 principal components in the PCA reference set from the number of sequences of each window of the first negative sample and the first sample to be detected after data amount correction, and obtain the number of sequences of each window in the first negative sample and the sample to be detected after PCA correction, respectively.

[0076] If a window contains maternal CNV, the sequence number of the window changes by about 50% compared with the window without CNV. Therefore, removing the interference of maternal CNV is very important for accurately detecting the copy number variation type of the fetus of a pregnant woman. In a preferred embodiment, the maternal CNV removal module is configured to use a hidden Markov chain to remove the maternal CNV in the sequence number after PCA correction in each window in the first sample to be tested.

[0077] In a third typical embodiment of the present application, a computer-readable storage medium is provided, the storage medium including a stored program, wherein when the program is executed, the device where the storage medium is located is controlled to execute the above-mentioned sliding window method for detecting prenatal fetal copy number abnormalities.

[0078] In a fourth typical embodiment of the present application, a processor is provided, and the processor is used to run a program, wherein the program runs the above-mentioned sliding window method for detecting prenatal fetal copy number abnormalities.

[0079] The present application is further described in detail below in conjunction with specific embodiments. These embodiments should not be construed as limiting the scope of protection claimed in the present application.

[0080] Example 1 (detection process such as Figure 1 (shown)

[0081] The method was tested using 200 negative samples, 22 T21 positive samples, and 15 DGS positive samples:

[0082] The above samples are samples of the Youxin'an project of Beijing Youxun Commercial Testing. T21 is trisomy 21, that is, Down syndrome, and DGS is Digeorge syndrome.

[0083] Among them, 100 negative samples were used as negative samples for the detection of this application, and the remaining 100 negative samples, 22 T21 positive samples, and 15 DGS positive samples were used as samples to be tested for the detection of this application.

[0084] 1. After preliminary quality control of the test samples and negative samples from the original offline data of the MGI-2000 platform using Trimmomatic-0.39 software, they were aligned to the human reference genome sequence (version hg19) using BWA 0.7.17, and Picard v2.20.3 was used to remove duplicate sequences, and Samtools v1.2 was used to extract the uniquely aligned base sequences. The chromosomes were divided into 100kb windows, with 50kb overlap between each window, and the number of uniquely aligned base sequences in each window was counted to generate a BAM file;

[0085] 2. Base correction. The number of unique base sequences aligned to each window is used to calculate the GC content (e.g. Figure 2 ) and regional mappability correction (as shown in Figure 3 shown);

[0086] 3. Data volume correction. Then the total data volume is set to 40M and allocated to each window according to the proportion of the base sequence in each window to the total data volume actually measured;

[0087] 4. PCA correction. The number of sequences in each window after data volume correction is subjected to PCA correction, and then the original content of the number of sequences in each window in the sample to be tested is calculated;

[0088] The specific process is as follows: the number of sequences corrected for the amount of data in the reference database samples (each window of negative samples greater than 10) is used to form a matrix, where each column of the matrix is ​​a sample and each row is a window. The matrix is ​​normalized and then SVD decomposition is performed to extract the first 10 principal components. The number of sequences corrected for the amount of data in each sample in the reference database is normalized and the first 10 principal components are removed. The resulting value is called the number of sequences corrected for this chromosome or this region by PCA.

[0089] 5. Using the number of sequences in each window after PCA correction, the hidden Markov model is used to calculate the posterior probability that each window in the sample to be tested carries the maternal CNV, and then the windows carrying the maternal CNV are eliminated (i.e., the probability that the maternal CNV status is negative is less than 0.99), and the number of sequences in each window in each sample that can be used for Z value calculation is obtained;

[0090] 6. CBS, HMM, fixed Z value and the sliding window method of this application were used to detect another 100 negative samples, 22 T21 samples and 15 DGS samples. The results are shown in Tables 1 and 2 below:

[0091] Among them, CBS specifically refers to: Circular Binary Segmentation; HMM specifically refers to: Hidden Markov Model; Fixed Z value specifically refers to: calculating the Z value for a fixed area (area to be detected).

[0092] The Z value of this application is calculated according to the following formula I:

[0093] Zscore=hh test -mean(hh) / sd(hh), formula I;

[0094] hh test : The average value of the number of window sequences of the sample in the area to be detected. mean(hh): The average value of the number of window sequences of all samples in the reference database in the area to be detected. sd(hh): The variance of the number of window sequences of all samples in the reference database in the area to be detected.

[0095] Construct a table with rows as Z-scores and columns as "copy number deletion" and "copy number duplication".

[0096] If the Z value is less than or equal to -3, the initial window to be detected is a copy number deletion window, the starting coordinates of the window to be detected are determined as the starting coordinates of the copy number deletion region, and the Z value is filled in the "Copy Number Deletion" column of the table; if the Z value is greater than or equal to 3, the initial window to be detected is a copy number duplication window, the starting coordinates of the window to be detected are determined as the starting coordinates of the copy number duplication region, and the Z value is filled in the "Copy Number Duplication" column of the table; if the Z value is less than 3 and greater than -3, the initial window to be detected is a normal window and does not carry copy number variation.

[0097] If the initial window to be detected is a normal window, the start and end coordinates of the area to be detected are both moved backward by one unit of length, and the Z value judgment of the subsequent single window to be judged is repeated until the copy number variation is detected or all windows are detected;

[0098] If the initial window to be detected is a copy number deletion window, the starting coordinates of the copy number deletion region remain unchanged, and the ending coordinates move backward by one unit length, and the Z value judgment of the subsequent single window to be judged is repeated until the Z value of the window to be detected is greater than the minimum value of the "copy number deletion" column in the table + 1, then the ending coordinates of the window to be detected are the ending coordinates of the copy number deletion region;

[0099] If the initial window to be detected is a copy number duplication window, the starting coordinates of the copy number duplication region remain unchanged, and the ending coordinates move backward by one unit length, and the Z value judgment of the subsequent single window to be judged is repeated until the Z value of the window to be detected is greater than the maximum value -1 of the "copy number duplication" column in the table, then the ending coordinates of the window to be detected are the ending coordinates of the copy number deletion region.

[0100] If the calculated absolute value of the Z value is greater than 3, the area is judged to be high risk, otherwise it is judged to be low risk.

[0101] Table 1 The positive detection rate of 22 cases of T21 positive and 15 cases of DGS positive

[0102] T21 DGS CBS 100% 86.70% HMM 95.20% 73.30% Fixed Z value 100% 66.70% Sliding Window Method 100% 100%

[0103] Table 2 Detection positive rate of 100 negative samples

[0104] T21 DGS CBS 6% 4% HMM 12% 13% Fixed Z value 2% 3% Sliding Window Method 0 0

[0105] As shown above, among several detection methods, the sliding window method of the present application has a positive detection rate of 100% for positive samples and a positive detection rate of 0% for negative samples.

[0106] From the above description, it can be seen that the above embodiments of the present invention achieve the following technical effects: Under the condition of less harm to pregnant women, the present application effectively improves the detection rate of positive samples and reduces the false positive rate by multi-level data correction of sequencing data of negative samples and samples to be tested. And the sliding window method based on Z value detection can improve the accuracy of fetal copy number variation detection, which is more sensitive and specific than fixed region Z value detection, CBS detection and HMM detection.

[0107] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for detecting fetal copy number abnormalities before birth, characterized in that: The method comprises: S1, dividing the genome sequencing data into windows of the same length, counting the number of reads of the sequencing data of the test sample and the negative sample in each window, and obtaining the original number of reads in each window in the first test sample and the first negative sample; S2, performing base correction, data amount correction and PCA correction on the original number of read segments in each window of the first sample to be tested and the first negative sample in turn, to obtain the number of sequences in each window of the first sample to be tested and the first negative sample after the PCA correction respectively; S3, removing the maternal CNV from the number of sequences in each window of the first sample to be tested after the PCA correction, and re-performing the data amount correction to obtain the corrected number of sequences in each window of the first sample to be tested; S4, calculating a Z value using the number of sequences in each window of the first negative sample after the PCA correction and the number of sequences in each window of the first sample to be tested after the correction, and determining the copy number variation type and variation region of the sample to be tested by a sliding window method using the Z value; Wherein, the basic correction includes correction of GC content and comparability.

2. The method according to claim 1, characterized in that The Z value is calculated using the following formula I; Zscore=hh test -mean(hh) / sd(hh), Formula I; Among them, hh test is the average value of the contents of all windows covered by the area to be detected in the first sample to be tested after correction; mean(hh) is the average value of the number of sequences of all windows covered by the region to be detected in the first negative sample after the PCA correction; sd(hh) is the variance of the number of sequences of all windows covered by the region to be detected in the first negative sample after the PCA correction.

3. The method according to claim 2, characterized in that The S4 includes: S41, constructing a table with the rows being the Z values ​​and the columns being the copy number states, wherein the copy number states include "copy number deletion" and "copy number duplication"; S42, using the calculated Z value to determine the copy number status of the initial window to be detected, and filling the Z value and the copy number status of the initial window to be detected into the table; S43, starting from the end coordinate of the initial window to be detected in the sample to be detected and ending at the end coordinate of the last window, sliding backward in units of the length of a window, performing the sliding window method judgment once each time sliding to one of the units, and obtaining the copy number variation type and variation region of the sample to be detected; The sliding window method comprises: using the calculated Z value to determine the copy number status of the window to be determined, and filling the Z value and the copy number status of the window to be determined into the table; The initial window to be detected is a window of the length of one unit after the end coordinates of the first window, each normal window, copy number deletion region or copy number duplication region in the sample to be tested.

4. The method according to claim 3, characterized in that The S42 includes: If the Z value is ≤-3, the initial window to be detected is a copy number deletion window, the starting coordinates of the window to be detected are determined as the starting coordinates of the copy number deletion region, and the Z value is filled in the "copy number deletion" column of the table; If the Z value is ≥ 3, the initial window to be detected is a copy number duplication window, the starting coordinates of the window to be detected are determined as the starting coordinates of the copy number duplication region, and the Z value is filled in the "copy number duplication" column of the table; If the Z value is less than 3 and greater than -3, the initial window to be detected is a normal window and does not carry copy number variation.

5. The method according to claim 4, characterized in that The S43 includes: If the initial window to be detected is a normal window, the start and end coordinates of the region to be detected are both moved backward by the length of the unit, and the operation of S42 is repeatedly performed until a copy number variation is detected or all windows are detected; If the initial window to be detected is a copy number deletion window, the starting coordinates of the copy number deletion region remain unchanged, and the ending coordinates move backward by the length of the unit, and the sliding window method judgment operation is repeatedly performed on the subsequent single window to be judged until the Z value of the window to be detected is greater than the minimum value of the "copy number deletion" column in the table + 1, then the ending coordinates of the window to be detected are the ending coordinates of the copy number deletion region; If the initial window to be detected is a copy number duplication window, the starting coordinates of the copy number duplication region remain unchanged, and the ending coordinates move backward by the length of the unit, and the sliding window method judgment operation is repeatedly performed on the subsequent single window to be judged until the Z value of the window to be detected is greater than the maximum value -1 of the "copy number duplication" column in the table, then the ending coordinates of the window to be detected are the ending coordinates of the copy number deletion region.

6. The method according to claim 1, characterized in that The sample to be tested and the negative sample include: peripheral blood of pregnant women; Preferably, the length of the window is 50-100 kb; Preferably, the basic correction comprises: using a local weighted regression method to perform the GC content correction and the comparability correction on the first sample to be tested and the first negative sample respectively, to obtain the number of sequences in each window of the first sample to be tested and the first negative sample after the basic correction; Preferably, the ratio of the number of sequences of each window after the basis correction to the number of sequences of all windows after the basis correction is calculated, and the total data volume is 40-50M, and the number of sequences of each window after the basis correction in the total data volume is calculated according to the ratio to obtain the number of sequences of each window after the data volume correction; Preferably, a hidden Markov chain is used to remove the maternal CNV in the number of sequences after the PCA correction in each window in the first sample to be tested.

7. The method according to claim 1, characterized in that The PCA correction includes: A data reference set is constructed using the number of sequences in each window of the first negative samples greater than 10 after correction for the data amount, and PCA training is performed on the reference set to form a PCA reference set; The number of sequences of each window of the first negative sample and the first sample to be detected after data amount correction is removed from the first 10 principal components in the PCA reference set to obtain the number of sequences of each window in the first negative sample and the sample to be detected after PCA correction.

8. A device for detecting abnormalities in fetal copy number before birth, characterized in that: The device comprises: A window segmentation module is configured to divide the genome sequencing data into windows of the same length, count the number of reads of the sequencing data of the test sample and the negative sample in each window, and obtain the number of original reads in each window of the first test sample and the first negative sample; A correction module is configured to sequentially perform basic correction, data amount correction and PCA correction on the number of original read segments in each window of the first sample to be tested and the first negative sample, and obtain the number of sequences in each window of the first sample to be tested and the first negative sample after the PCA correction respectively; a maternal CNV removal module, configured to remove the maternal CNV from the number of sequences in each window of the first sample to be tested after the PCA correction, and re-perform the data amount correction to obtain the corrected number of sequences in each window of the first sample to be tested; A sliding window detection module is configured to calculate a Z value using the number of sequences in each window of the first negative sample after the PCA correction and the number of sequences in each window of the first sample to be tested after the correction, and use the Z value to determine the copy number variation type and variation region of the sample to be tested by a sliding window method; Wherein, the basic correction includes correction of GC content and comparability.

9. The device according to claim 8, characterized in that The sliding window detection module includes a Z value calculation unit: The Z value calculation unit is configured to calculate the Z value using the following formula I; Zscore=hh test -mean(hh) / sd(hh), Formula I; Among them, hh test is the average value of the contents of all windows covered by the area to be detected in the first sample to be tested after correction; mean(hh) is the average value of the number of sequences of all windows covered by the region to be detected in the first negative sample after the PCA correction; sd(hh) is the variance of the number of sequences of all windows covered by the region to be detected in the first negative sample after the PCA correction.

10. The device according to claim 9, characterized in that The sliding window detection module comprises: A construct table unit is configured to construct a table with rows being the Z values ​​and columns being copy number states, wherein the copy number states include "copy number deletion" and "copy number duplication"; The initial window to be detected unit is configured to use the calculated Z value to determine the copy number status of the initial window to be detected, and fill the Z value and the copy number status of the window to be detected into the table; The output copy number variation unit is configured to start from the end coordinates of the first window in each of the regions to be detected in the sample to be detected and end at the end coordinates of the last window, and slide backward in sequence in units of the length of a window, and perform the sliding window method judgment once each time sliding to a unit to obtain the copy number variation type and variation region of the sample to be detected; The sliding window method comprises: using the calculated Z value to determine the copy number status of the window to be determined, and filling the Z value and the copy number status of the window to be determined into the table; The initial window to be detected is a window of the length of one unit after the end coordinates of the first window, each normal window, copy number deletion region or copy number duplication region in the sample to be tested.

11. The device according to claim 10, characterized in that The detecting initial window unit to be tested comprises: a copy number deletion judgment unit, configured to determine that if the Z value is ≤ -3, the initial window to be detected is a copy number deletion window, the starting coordinates of the window to be detected are determined as the starting coordinates of the copy number deletion region, and the Z value is filled in the "copy number deletion" column of the table; a copy number duplication judgment unit, configured to determine that if the Z value is ≥ 3, the initial window to be detected is a copy number duplication window, the starting coordinates of the window to be detected are determined as the starting coordinates of the copy number duplication region, and the Z value is filled in the "copy number duplication" column of the table; The normal judgment unit is configured to determine that if the Z value is less than 3 and greater than -3, the initial window to be detected is a normal window and does not carry a copy number variation.

12. The device according to claim 11, characterized in that The output copy number variation unit comprises: The copy number normal unit is set such that if the initial window to be detected is a normal window, the start and end coordinates of the area to be detected are both moved backward by the length of the unit, and the operation of detecting the initial window unit to be detected is repeatedly performed until the copy number variation is detected or all windows are detected; The copy number deletion unit is configured such that if the initial window to be detected is a copy number deletion window, the starting coordinates of the copy number deletion region remain unchanged, the ending coordinates move backward by the length of the unit, and the sliding window method judgment operation is repeatedly performed on the subsequent single window to be judged until the Z value of the window to be detected is greater than the minimum value of the "copy number deletion" column in the table + 1, then the ending coordinates of the window to be detected are the ending coordinates of the copy number deletion region; The copy number duplication unit is set such that if the initial window to be detected is a copy number duplication window, the starting coordinates of the copy number duplication region remain unchanged, the ending coordinates move backward by the length of the unit, and the sliding window method judgment operation is repeatedly performed on the subsequent single window to be judged until the Z value of the window to be detected is greater than the maximum value -1 of the "copy number duplication" column in the table, then the ending coordinates of the window to be detected are the ending coordinates of the copy number deletion region.

13. The device according to claim 8, characterized in that The sample to be tested and the negative sample include: peripheral blood of pregnant women; Preferably, the length of the window is 50-100 kb; Preferably, the correction module comprises: a basic correction unit, configured to perform the GC content correction and the comparability correction on the first sample to be tested and the first negative sample in sequence using a local weighted regression method, to obtain the number of sequences in each window of the first sample to be tested and the first negative sample after the basic correction; Preferably, the correction module further comprises: a data amount correction unit, configured to calculate the ratio of the number of sequences of each window after the basis correction to the number of sequences of all windows after the basis correction, and to calculate the number of sequences of each window after the basis correction in the total data amount according to the ratio, so as to obtain the number of sequences of each window after the data amount correction, when the total data amount is 40-50M; Preferably, a hidden Markov chain is used to remove the maternal CNV in the number of sequences after the PCA correction in each window in the first sample to be tested.

14. The device according to claim 8, characterized in that The correction module further includes: a PCA correction unit, which is configured to construct a data reference set using the number of sequences of each window of the first negative samples greater than 10 after data amount correction, and perform PCA training on the reference set to form a PCA reference set; The number of sequences of each window of the first negative sample and the first sample to be detected after data amount correction is removed from the first 10 principal components in the PCA reference set to obtain the number of sequences of each window in the first negative sample and the sample to be detected after PCA correction.

15. A computer-readable storage medium, characterized in that: The storage medium includes a stored program, wherein when the program is executed, the device where the storage medium is located is controlled to execute the method for detecting prenatal fetal copy number abnormality according to any one of claims 1 to 7.

16. A processor, characterized in that: The processor is used to run a program, wherein the program runs the method for detecting prenatal fetal copy number abnormalities according to any one of claims 1 to 7.