Micro-satellite instability detection method and apparatus based on next-generation sequencing technology
By selecting specific MSI sites and constructing baseline samples in next-generation sequencing technology, and combining Spearman rank correlation coefficient judgment, the problems of accuracy and control sample requirements in existing technologies for microsatellite instability detection have been solved, achieving high accuracy and low tumor purity for microsatellite instability detection.
Patent Information
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- ZHENYUE BIOTECHNOLOGY JIANGSU CO LTD
- Filing Date
- 2025-01-10
- Publication Date
- 2026-07-16
AI Technical Summary
Existing microsatellite instability detection technologies suffer from problems such as high requirements for the purity of tumor tissue samples, the need for control samples, and insufficient detection accuracy. In particular, in next-generation sequencing technology, there is a lack of high-accuracy detection algorithms that are applicable to both hybridization capture and amplicon sequencing data.
A microsatellite instability detection method based on next-generation sequencing technology was adopted. By selecting MSI sites with repetitive sequence lengths between 10-15 bp, capture probes and/or amplicon primers were designed, the ratio of deleted and normal fragments was calculated, a baseline sample was constructed, the Spearman rank correlation coefficient was used to determine the sample status, and the baseline was adjusted to improve accuracy when anomalies were detected.
It achieves high-accuracy microsatellite instability detection without the need for control samples, maintains high specificity and sensitivity under conditions of low tumor purity and low sequencing data volume, reduces false positive rate, and is suitable for hybridization capture and amplicon sequencing data.
Smart Images
Figure CN2025071733_16072026_PF_FP_ABST
Abstract
Description
A method and apparatus for detecting microsatellite instability based on next-generation sequencing technology Technical Field
[0001] This invention belongs to the field of biological detection technology, specifically relating to a microsatellite instability detection method and device based on next-generation sequencing technology. Background Technology
[0002] The current gold standard for detecting microsatellite instability (MSI) primarily uses PCR to determine the MSI status of a sample. This method uses fluorescently labeled primers and capillary electrophoresis to determine the fragment length polymorphisms at five microsatellite loci: NR-21, NR-24, BAT-25, BAT-26, and MONO-27. By comparing the polymorphisms at these five microsatellite loci with those at normal control samples, the presence of instability at these loci can be confirmed. If two or more of the five microsatellite loci show microsatellite instability, it is classified as highly unstable (MSI-H); otherwise, it is classified as low-instability (MSI-L) or stable (MSS).
[0003] Besides PCR, next-generation sequencing methods based on hybridization capture and amplicon sequencing for MSI detection have developed rapidly, with various methods and tools published and applied in clinical testing and scientific research. However, each method has its own shortcomings and limitations, specifically:
[0004] 1) PCR testing is currently the gold standard for MSI detection, offering high accuracy, but it has high requirements for sample quality. PCR requires simultaneous testing of both tumor tissue samples and control samples, and the purity of the tumor tissue samples is also critical.
[0005] 2) MSI detection based on next-generation sequencing technology has low requirements for tumor purity and does not require control samples. These methods can simultaneously perform microsatellite instability detection and other detections, but the accuracy of the results still needs to be improved.
[0006] Furthermore, due to the different data characteristics of hybridization capture and amplicon sequencing, there is currently no second-generation sequencing MSI detection algorithm that can be used simultaneously for hybridization capture and amplicon sequencing data. Summary of the Invention
[0007] The present invention aims to at least partially solve one of the technical problems in the aforementioned related technologies.
[0008] Therefore, the purpose of this invention is to provide a microsatellite instability detection method and device based on next-generation sequencing technology, which has high stability and high accuracy, and does not require control samples, and can be applied to the detection of microsatellite instability in hybridization capture and amplicon sequencing data simultaneously.
[0009] To solve the above-mentioned technical problems, the present invention is implemented as follows:
[0010] This invention provides a method for detecting microsatellite instability based on next-generation sequencing technology, the method comprising:
[0011] Select a set of MSI sites with repeat sequence lengths between 10-15 bp, and design capture probes and / or amplicon primers based on any combination of the MSI sites.
[0012] Select several samples, and for each sample, calculate the deletion ratio and reference ratio at each MSI site.
[0013] Several samples that were determined to be in MSI-L or MSS status by PCR were selected from the above samples as baseline samples. Different depth levels were set for each baseline sample, and a baseline was constructed based on each depth level.
[0014] Based on the established baseline, MSI status detection is performed on each selected sample;
[0015] Select samples with an MSI status detection result of MSI-H, and determine whether they are anomalous samples by calculating the Spearman rank correlation coefficient; if they are determined to be anomalous samples, adjust the baseline of the sample and re-determine the MSI status.
[0016] In addition, the microsatellite instability detection method based on next-generation sequencing technology according to the present invention may also have the following additional technical features:
[0017] In some implementations, the calculation of the proportion of missing segments and the proportion of normal segments in the sample includes:
[0018] a) Align the sequenced base sequence to the human genome reference sequence and extract sequencing base sequences that completely cover the region where the MSI site is located and at least two bases on each side;
[0019] b) Filter the sequencing data of the captured sequencing samples;
[0020] c) Remove duplicate sequences from the filtered sequencing base sequences of the captured sequencing samples;
[0021] d) For each remaining sequencing base sequence, count the repeat sequence length at the current MSI site. Based on the relative relationship between the repeat sequence length of the sequencing base sequence and the human genome reference sequence, classify the sequencing base sequences. Count the number of sequencing base sequences in each category and the total number of sequencing base sequences, and calculate the proportion of deleted fragments and the proportion of normal fragments.
[0022] In some implementations, filtering the sequencing data of the captured sequencing samples includes:
[0023] Extract three bases from each sequencing base sequence on each side of the MSI site; for MSI sites where the edges of the reads cover less than three bases, extract only two bases; if the sequencing base sequence at these positions is not completely consistent with the human genome reference sequence, filter out that sequencing base sequence.
[0024] Extract the sequence corresponding to the MSI site region from each sequencing base sequence; if there are two or more bases in the sequence that are inconsistent with the reference sequence, filter the sequencing base sequence; if there is one base in the sequence that is inconsistent with the human genome reference sequence and the sequence happens to be at the start or end position of the MSI site, filter out the sequencing base sequence.
[0025] In some implementations, the process of deduplicating the sequenced bases after filtering the captured sequencing samples includes:
[0026] For the captured and sequenced samples, the origin of the original DNA molecular fragment of the sequenced base sequence is determined according to the start and end sites of each sequenced base sequence; the sequenced base sequence is either a single-end sequenced base sequence or a paired-end sequenced base sequence;
[0027] For repeated sequencing base sequences originating from the same original DNA molecule fragment, one base sequence is randomly selected from the repeated sequencing base sequences and retained.
[0028] In some implementations, the process of building a baseline includes:
[0029] a) Calculate the proportion of missing fragments and the proportion of normal fragments for each baseline sample at the corresponding depth level of each MSI site;
[0030] b) Calculate the baseline for each MSI site at its depth level; this includes:
[0031] Calculate the baseline mean and standard deviation of the proportion of missing fragments in the baseline sample for each MSI locus;
[0032] Calculate the baseline mean and standard deviation of the proportion of normal fragments in the baseline sample for each MSI locus.
[0033] In some implementations, the MSI state detection for each selected sample, based on the established baseline, includes:
[0034] a) Calculate the proportion of missing fragments and the proportion of normal fragments at each MSI site for each sample;
[0035] b) Calculate the depth of each MSI site and match a baseline of equal depth level based on that depth;
[0036] c) Calculate the Z-score of the proportion of missing fragments and the proportion of normal fragments at each MSI site relative to the baseline;
[0037] d) Determine the status of the sample at each MSI site based on the calculated depth and Z_score;
[0038] The determination result is: site quality control unqualified, site unstable, or site stable;
[0039] e) Based on the overall status of all MSI sites, calculate the MSI score and count the number of sites that pass quality control. Determine the MSI status of the sample based on the statistical results.
[0040] In some implementations, the Spearman rank correlation coefficient is the Spearman rank correlation coefficient between the proportion of missing segments and the mean of the baseline proportion of missing segments;
[0041] Set an outlier threshold for the Spearman rank correlation coefficient. When the Spearman rank correlation coefficient is greater than or equal to the outlier threshold, the corresponding sample is judged as an outlier sample; otherwise, the corresponding sample is judged as normal.
[0042] In some of these implementations, the number of MSI sites, all with repeat sequence lengths between 10 and 15 bp, is 74.
[0043] In some implementations, adjusting the baseline of the abnormal sample and re-evaluating the MSI status includes: adjusting the baseline matched at each site of the sample, re-evaluating the site status based on the adjusted baseline, combining the evaluation results of all sites to obtain a new MSI status for the sample, and replacing the original MSI status with the new MSI status.
[0044] The baseline adjustment process includes: adjusting the baseline mean of the missing fragment ratio to minimize the sum of squared Z_scores of stable sites in the outlier samples; recalculating the Z_score of all sites based on the adjusted baseline mean of the missing fragment ratio and updating the site status; repeating the above steps until the status of all sites no longer changes.
[0045] This invention also provides a microsatellite instability detection device based on next-generation sequencing technology, including a processor and a memory. The memory stores a software program, and when the processor runs the software program, it can implement the microsatellite instability detection method based on next-generation sequencing technology as described above.
[0046] Compared with the prior art, the present invention has at least the following beneficial effects:
[0047] In this embodiment of the invention, the microsatellite instability detection method based on next-generation sequencing technology provides a set of 74 MSI sites, all of which have repeat sequence lengths between 10 and 15 bp. Compared with sites with repeat sequence lengths greater than 15 bp, the 74 sites selected in this invention can achieve higher alignment accuracy when aligned to the reference genome.
[0048] In this embodiment of the invention, the microsatellite instability detection method based on next-generation sequencing technology provides a method that uses any combination of 74 sites. Based on the distribution of repetitive sequence lengths at each site for any sample, the deletion ratio and reference ratio at each site can be calculated for the sample. This method is used to construct the MSI baseline and detect the MSI of the sample to be tested.
[0049] In this embodiment of the invention, the microsatellite instability detection method based on next-generation sequencing technology filters the next-generation sequencing data according to specific rules before calculating the deletion ratio and reference ratio, so as to reduce the impact of alignment errors on the distribution of repeat sequence lengths at microsatellite loci and obtain more accurate MSI detection results.
[0050] In this embodiment of the invention, the microsatellite instability detection method based on next-generation sequencing technology introduces a reference ratio as an auxiliary reference. Compared with using the deletion ratio alone for detection, it can identify some repetitive sequence length distribution shifts caused by experimental and sequencing errors, thereby further improving the detection accuracy.
[0051] In this embodiment of the invention, the microsatellite instability detection method based on next-generation sequencing technology uses a set of MSS / MSI-L samples. Based on the deletion ratio and reference ratio of this set of samples at each MSI site, an MSI baseline can be constructed. After the baseline is constructed, it can be used to perform MSI detection on any sample to be tested without providing a corresponding control sample for each sample.
[0052] In this embodiment of the invention, the microsatellite instability detection method based on next-generation sequencing technology is provided to determine whether a sample is abnormal. If the sample is abnormal, the MSI baseline is adjusted to adapt to the abnormal sample. This can reduce the false differences between the sample to be tested and the MSI baseline caused by improper sample preservation, unstable library construction experimental procedures, or sequencing abnormalities, and reduce the false positive rate of MSI detection (the proportion of samples that are detected as MSI-H but are actually MSI-L or MSS).
[0053] In this embodiment of the invention, the microsatellite instability detection method based on next-generation sequencing technology determines the stability of each MSI site by comparing the differences in deletion ratio and reference ratio between the test sample and the baseline at each MSI site. It then integrates the stability of multiple sites to determine the MSI status of the test sample, which has low requirements for tumor purity. Under any amount of data, this invention can maintain stable high specificity in MSS / MSI-L samples. Even when the amount of sequencing data is reduced, it can still obtain high sensitivity in samples with tumor purity as low as 5%-10%.
[0054] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description
[0055] Figure 1 is a schematic diagram of the detection limit of a microsatellite instability detection method based on second-generation sequencing technology disclosed in an embodiment of the present invention in captured sequencing data. Detailed Implementation
[0056] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0057] The embodiments of the present invention will be described in detail below with reference to the accompanying drawings and specific examples and application scenarios.
[0058] Please refer to Figure 1. In some embodiments of the present invention, a microsatellite instability (MSI) detection method based on next-generation sequencing technology is provided. This method can be used for capture sequencing or amplicon sequencing, and can accurately detect the MSI status of tumor tissue samples or body fluid samples when the proportion of tumor cells is low and no control samples are available, thus indicating the treatment options available to the patient. The method employs the following technical solution:
[0059] Step 1. Select 74 MSI sites for designing capture probes or amplicon primers. Any combination of the 74 MSI sites can be used when designing capture probes or amplicon primers. The specific locations of these sites on the human reference genome hg19 are shown in Table 1.
[0060] Table 1. Repeat sequence length information and detailed location of 74 MSI sites
[0061] Step 2. Select S out of the 74 MSI loci. For each sample, calculate the deletion ratio and reference ratio at each MSI locus.
[0062] a) Align the sequenced base sequence to the human genome reference sequence and extract reads (sequencing base sequences) that completely cover the region where the MSI site is located and at least two bases on each side.
[0063] b) Filter the sequencing data of the captured sequencing samples (skip this step for amplicon sequencing samples):
[0064] ① Extract a sequence of three bases from each flanking position of the MSI site on each read. If the flanking positions of the MSI site are at the edge of the read and cover less than three bases, extract only two bases. If the base sequence of the read at these positions is not completely consistent with the human genome reference sequence, filter the read.
[0065] ② Extract the sequence corresponding to the MSI site region from each read. If there are two or more bases in this sequence that are inconsistent with the reference sequence, filter the read; if there is one base in this sequence that is inconsistent with the human genome reference sequence, and this sequence happens to be at the start or end position of the MSI site, filter the read.
[0066] c) Remove duplicates from the filtered reads of the captured sequencing samples (skip this step for amplicon sequencing samples):
[0067] For captured sequencing samples, the origin of the original DNA molecular fragment of each read is determined according to the start and end sites of each sequencing read (single-end sequencing) or each pair of reads (paired-end sequencing).
[0068] For repeated reads originating from the same original molecular fragment, one read is randomly selected from the repeated reads and retained.
[0069] d) For each remaining read, calculate the length of the repeat sequence at the current MSI locus. Based on the relative length of the repeat sequence to the human genome reference sequence, classify the reads as follows: (1) consistent length: normal (reference read); (2) shorter than the reference sequence: deletion read; (3) longer than the reference sequence: insertion read. Calculate the ratio of reference reads to all reads at the current locus (reference ratio) and the ratio of deletion reads to all reads (deletion ratio).
[0070] Step 3. Select N samples with MSI-L or MSS status as baseline samples. For the selected S MSI sites, set different depth levels {d1, d2, ..., d...}. D Baselines were constructed for each depth level d. The MSI state of the baseline samples selected here was pre-determined by the PCR method.
[0071] a) For each of the S sites s, filter and deduplicate the reads of each baseline sample n according to the methods described in steps 2.a-2.c (this step is skipped for amplicon sequencing), and randomly select d reads from the retained reads.
[0072] b) Using the method described in step 2.d, calculate the deletion ratio and reference ratio of each baseline sample n at depth level d at site s, denoted as del_ratio, respectively. s,n,d and ref_ratio s,n,d .
[0073] c) Calculate the baseline at depth level d at each site s. s,d This includes:
[0074] Mean and standard deviation of deletion ratio of site s in the baseline sample: mean_del s,d and SD_del s,d ;
[0075] Baseline mean and standard deviation of the reference ratio of loci s in the baseline sample: mean_ref s,d and SD_ref s,d .
[0076] Step 4. Detect the MSI state of the sample i to be tested:
[0077] a) Calculate the deletion ratio at each site s among the S sites of the sample, using the method described in section 2: {del_ratio1, del_ratio2, ..., del_ratio...} S}; and reference ratio: {ref_ratio1,ref_ratio2,...,ref_ratio S};
[0078] b) Calculate the depth d at each site s. s According to d s Match a baseline of equal depth level to site s. The matching method is: select and d s The d with the smallest difference;
[0079] c) Calculate the deletion ratio (del_ratio) of the sample at each site. s ) and reference ratio (ref_ratio) s Relative to baseline Z-score:
[0080] d) According to d s Z_del s and Z_ref s The state T of the sample to be tested at site s s Make a judgment:
[0081] If d s <d_cutoff: Site quality control failed;
[0082] If d s ≥d_cutoff and simultaneously satisfy Z_del s ≥Z_cutoff and Z_del s >-Z_ref s Site instability;
[0083] If d s If the site is ≥d_cutoff and does not meet the conditions in the previous condition, the site is stable;
[0084] Where d_cutoff is the quality control threshold for MSI site depth, and Z_cutoff is the Z-score threshold for MSI sites.
[0085] e) Obtaining T at each site sThen, considering the status of all sites, the MSI score is calculated and the number of sites that pass quality control is counted. The MSI status of the sample to be tested is then determined for the first time.
[0086] Count the number of unstable loci (c_unstable) and the number of stable loci (c_stable); the number of loci that pass quality control is equal to the number of unstable loci (c_unstable + c_stable).
[0087] If passing_loci ≥ loci_cutoff and msi_score < score_cutoff: the sample to be tested is MSS / MSI-L;
[0088] If passing_loci ≥ loci_cutoff and msi_score ≥ score_cutoff: the sample to be tested is MSI-H;
[0089] If passing_loci < loci_cutoff and c_unstable ≥ c_cutoff: the sample to be tested is MSI-H;
[0090] If passing_loci < loci_cutoff and c_unstable < c_cutoff: the quality control of the sample to be tested is not satisfactory (QNS);
[0091] Where loci_cutoff is the threshold for qualified quality control sites; score_cutoff is the MSI score threshold when the number of quality control sites is greater than or equal to loci_cutoff; and c_cutoff is the threshold for the number of unstable sites when the number of quality control sites is less than loci_cutoff.
[0092] Step 5. If the sample was determined to be MSI-H in Step 4, you can choose to calculate ρ for the sample to be tested to determine whether the sample is an outlier. If you choose to calculate, the calculation and usage of ρ are as follows:
[0093] a) Calculate the deletion ratio ({mean_del1,mean_del2,…,mean_del...) of the sample i to be tested. S}) and the mean of the baseline deletion ratio ({del_ratio) 1,i del_ratio 2,i ,…,del_ratio S,iThe Spearman rank correlation coefficient ρ of the sample is used to determine the deletion ratio between the sample and the baseline at the S MSI sites. A higher value indicates a stronger correlation between the deletion ratio of the sample and the baseline at the S MSI sites.
[0094] b) ρ_cutoff is the outlier threshold for ρ. Compare ρ with ρ_cutoff:
[0095] If ρ is greater than or equal to ρ_cutoff, it indicates that the deletion ratio of the sample under test and the baseline at the S MSI sites is strongly correlated. The sample may have shown that the deletion ratio increases by the same magnitude at all MSI sites. This situation is relatively rare in real MSI-H samples, and the sample is more likely to have experimental or sequencing abnormalities. These samples are marked as abnormal samples.
[0096] If ρ is less than ρ_cutoff, it indicates that the correlation between the test sample and the baseline at the S MSI sites is relatively weak, which is consistent with the determination status of MSI-H.
[0097] Step 6. If the operation in Step 5 was performed, and the sample to be tested was marked as an anomalous sample, then the baseline matched for that sample at each site s is... Make adjustments and base them on the adjusted baseline. The site status is reassessed. Finally, the new results from all site assessments are combined to obtain the MSI status of the sample, replacing the result obtained in step 4.
[0098] a) to of Adjustments were made, and the adjusted baseline is: The adjustment method is as follows:
[0099] ① Obtain the set I of MSI sites that are currently identified as stable;
[0100] ② For each stable site i in I, calculate Among them, del_ratio i It is the deletion ratio of the sample at site i. and is the mean and standard deviation of the baseline deletion ratio matched at site i, and a and b are the parameters that need to be estimated;
[0101] ③ Calculate the Z_del for all sites in I. i ′Sum of squares: Iterate through all possible combinations of parameters a and b to find the one that minimizes... Find the optimal solution and obtain the estimated values of a and b.
[0102] b) Using the estimated parameters a and b, calculate the values of all S MSI sites. use replace get
[0103] c) Repeat steps 4.c-4.e, based on... Calculate the MSI score of the sample under test after this round of correction: msi_score′;
[0104] d) Compare the changes in the MSI score before and after this round of correction:
[0105] If msi_score′ < msi_score, update the MSI stable site set I and continue repeating steps 6.a-6.c;
[0106] If msi_score′≥msi_score, output the current msi_score′.
[0107] e) Based on the standard described in step 4.e, determine the MSI status of the sample to be tested according to the corrected msi_score′.
[0108] Example 1: Hybridization capture sequencing data was used to verify the performance of the MSI detection method described in steps 1-4 of this invention.
[0109] 1. Sample preparation and sample information
[0110] a) Baseline clinical samples: A total of 67 MSS samples were obtained from non-small cell lung cancer patients.
[0111] b) Cell line simulation validation samples:
[0112] Cell lines 22RV1, RL95, and GM12878 were used for pooling. GM12878 had an MSI (Microsatellite Instability) status of MSS (Microsatellite Stable), while 22RV1 and RL95 had an MSI status of MSI-H (High Microsatellite Instability). 22RV1 or RL95 was incorporated into the GM12878 sample at different proportions, resulting in five pools with 22RV1 mass percentages of 20%, 10%, 5%, 2.5%, and 1.25%, and five pools with RL95 mass percentages of 20%, 10%, 5%, 2.5%, and 1.25%, respectively. These five pools were used as simulated MSI-H tissue samples with tumor purity of 20%, 10%, 5%, 2.5%, and 1.25%, respectively.
[0113] Ten replicate library construction experiments were performed on the MSI-H tissue simulation samples and the GM12878 samples without pooling, resulting in a total of 110 libraries. Specific library information and quantities are shown in Table 2.
[0114] Table 2. Detailed information on the capture sequencing cell line simulation validation samples
[0115] c) Clinical validation samples: a total of 181 cases, including 148 MSS (microsatellite stable) or MSI-L (low microsatellite instability) samples and 33 MSI-H (high microsatellite instability) samples.
[0116] 2. Acquire captured sequencing data and perform data preprocessing.
[0117] Capture probes were designed for 74 MSI sites (Table 1). Library construction experiments were performed on 67 baseline clinical samples, 110 cell line simulation validation samples, and 181 clinical validation samples, and hybridization capture was performed using the capture probes. The captured libraries were sequenced end-to-end using the DNBSEQ-T7 sequencing platform to obtain raw sequencing data (bcl files). The bcl files were split using bcl2fastq to obtain paired-end sequencing FastQ data for each sample. The sequencing data volume for each sample was approximately 2GB, with a depth of approximately 4000x before deduplication.
[0118] In this embodiment, for each sample's FastQ data, Trimmomatic is used to remove adapter sequences and low-sequencing-quality sequences from each read. The remaining sequences of the reads are then aligned to the human reference genome hg19 using BWA to obtain the original BAM file. Reads aligned to MSI sites in the original BAM file are extracted and re-aligned using GATK Realigner to obtain more accurate alignment positions, which are then stored in the preprocessed BAM file.
[0119] 3. MSI baseline was constructed using 67 baseline samples.
[0120] a) For each MSI site, extract reads from the preprocessed bam of each sample that completely cover the region where the site is located and at least two bases on each side, following the method described in steps 2.a-2.c, and then filter and remove duplicates.
[0121] b) Set different depth levels: 100x, 150x, 200x, 250x, 300x, 400x, 500x, 600x, 800x, 1000x, 1200x, 1500x, 2000x, 2500x, 3000x, 3500x, 4000x. For each depth level:
[0122] First, extract the corresponding number of reads from the deduplicated reads at site s for each sample, and calculate the deletion ratio and reference ratio of the 67 samples at site s at this depth level.
[0123] Then, the mean and standard deviation of the deletion ratio and the standard deviation of the reference ratio of the 67 samples at the corresponding depth level were calculated and recorded as the baseline at that depth level.
[0124] c) Merge the baselines at different depth levels and save them as a single baseline file.
[0125] 4. Use the MSI baseline obtained in the previous step to detect and verify the MSI status of the sample.
[0126] a) For each site s of each sample to be tested, perform the following steps:
[0127] ① Following the method described in step 2.a, extract reads from the preprocessed bam of each sample that fully cover the region of the site and at least two bases long on each side, and then filter and remove duplicates;
[0128] ② Following the method described in step 4.a, calculate the deletion ratio and reference ratio of the test sample at site s based on the deduplicated reads;
[0129] ③ Following the method described in step 4.b, calculate the depth of the sample to be tested at site s based on the extracted reads, and match a suitable baseline for the sample according to the depth;
[0130] ④ Using the formula in step 4.c, calculate the Z_del of the test sample at site s based on the deletion ratio and reference ratio of the test sample and the matched baseline. s and Z_ref s ;
[0131] ⑤ Set d_cutoff to 100 and Z_cutoff to 4, and determine the state of the test sample site s according to the rules in step 4.d.
[0132] b) For each sample to be tested, count the number of stable sites and unstable sites at the 74 MSI sites, and calculate the MSI score. Set loci_cutoff to 40, score_cutoff to 0.1, and c_cutoff to 6, and determine the MSI status of the sample according to the rules in step 4.e.
[0133] c) Statistical analysis of MSI detection results in cell line simulation validation samples:
[0134] In this embodiment, the sensitivity in replicates with different mixing ratios of 22RV1 and RL95, and the specificity in replicates of GM12878 are shown in Table 3. Sensitivity refers to the proportion of genuine MSI-H samples that are correctly detected; specificity refers to the proportion of genuine MSS / MSI-L samples that are correctly detected.
[0135] Table 3. Detection of samples from the capture sequencing cell line simulation validation
[0136] In this embodiment, Probit regression was used to fit the sensitivity of simulated validation samples of cell lines with different tumor cell percentages (as shown in Figure 1) to obtain the limit of detection (LOD) in the captured sequencing data of this embodiment: when 95% sensitivity is required, the LOD of this embodiment is 2.61%, proving that this embodiment can be used for the detection of samples with low tumor purity. In Figure 1, the x-axis represents the tumor cell percentage, and the y-axis represents the detection sensitivity of the present invention in captured sequencing samples with the corresponding tumor cell percentage; the solid line in the figure represents the relationship between y and x fitted by Probit regression based on the sensitivity in simulated cell line samples with different tumor cell percentages; through the intersection of the dashed line y = 0.95 and the solid line, it can be seen that the detection limit of the present invention is 2.61% when the sensitivity is 95%, that is: when the tumor purity is 2.61%, the present invention can obtain 95% sensitivity.
[0137] In this embodiment, the detection results of MSI in clinical validation samples were statistically analyzed: the sensitivity and specificity of this embodiment in clinical validation samples were 100%. Sensitivity refers to the proportion of real MSI-H samples that were correctly detected; specificity refers to the proportion of real MSS / MSI-L samples that were correctly detected. This demonstrates that this embodiment can achieve high performance in clinical samples.
[0138] 5. Verify detection performance under different data volumes
[0139] a) Reduce the data volume of cell line simulation and clinical validation samples to verify the performance of this invention in low-volume sequencing data capture. Randomly select 75%, 50%, 25%, 20%, and 10% of reads from the paired-end sequencing FASTQ data of each sample. Then, use Trimmomatic to remove adapter sequences and low-sequencing-quality sequences from each read. Align the remaining sequences of the reads to the human reference genome hg19 using BWA to obtain the original BAM file. Extract the reads aligned to MSI sites from the original BAM file and re-align these reads using GATK Realigner to obtain more accurate alignment positions. Store these reads in the preprocessed BAM file.
[0140] b) Following the MSI detection method for validation samples with the original data volume in this embodiment, MSI status was detected for validation samples obtained after randomly selecting different data volumes. The performance of cell line simulation validation samples under different data volumes was statistically analyzed (Table 4) and the performance of clinical validation samples (Table 5):
[0141] In this embodiment, with the amount of sequencing data reduced to 10%, only one case of QNS occurred, and the remaining validation samples were able to meet the quality control conditions and obtain test results, proving that this embodiment has high stability.
[0142] With a data volume of 10%, this embodiment can maintain 100% specificity. Although the sensitivity is reduced, the detection limit can still reach 5.72%, which proves that this embodiment has high detection performance with low sequencing data volume.
[0143] Table 4. Detection limits of captured sequencing cell lines in simulated validation samples under different data volumes
[0144] Table 5. Detection performance of capture sequencing in clinical validation samples under different data volumes
[0145] Example 2: Amplicon sequencing data, used to verify the effectiveness of the present invention in amplicon sequencing and the abnormal sample correction method described in steps 5-6.
[0146] 1. Sample preparation and sample information
[0147] Baseline clinical samples: 35 MSS / MSI-L samples.
[0148] Clinical validation samples: a total of 103 cases, including 84 MSS / MSI-L samples and 19 MSI-H samples.
[0149] 2. Acquire amplicon sequencing data and perform data preprocessing.
[0150] Amplicon primers were designed for 68 MSI sites (Table 6). Library construction experiments were performed on 35 baseline clinical samples and 103 clinical validation samples, and amplicon primers were used for enrichment. Single-end sequencing of the amplicon libraries was performed using the Miniseq sequencing platform to obtain raw bcl files. The bcl files were split using bcl2fastq to obtain paired-end sequencing FastQ data for each sample.
[0151] For each sample's FastQ data, the adapter sequences of each read were removed using Trimmomatic, and the remaining sequences of the reads were aligned to the human reference genome hg19 using BWA to obtain the original BAM file. Reads aligned to MSI sites in the original BAM file were extracted, and these reads were re-aligned using GATK Realigner to obtain more accurate alignment positions, which were then stored in the preprocessed BAM file.
[0152] Table 6. 68 MSI sites used for designing amplicon primers
[0153] 3. MSI baseline was constructed using 35 baseline samples.
[0154] a) For each MSI site, extract reads from the preprocessed bam of each sample that fully cover the region where the site is located and at least two bases on each side, following the method described in step 2.a.
[0155] b) Set different depth levels: 100x, 150x, 200x, 250x, 300x, 400x, 500x, 600x, 800x, 1000x, 1200x, 1500x, 2000x, 2250x, 2500x, 2750x, 3000x, 3250x, 3500x, 3750x, 4000x, 4500x, 5000x, 5500x, 6000x, 6500x, 7000x. For each depth level:
[0156] First, extract the corresponding number of reads from the reads at site s for each sample, and calculate the deletion ratio and reference ratio of the 35 samples at site s at this depth level.
[0157] Then, the mean and standard deviation of the deletion ratio and the standard deviation of the reference ratio of the 35 samples at the corresponding depth level are calculated and recorded as the baseline at that depth level.
[0158] c) Merge the baselines at different depth levels and save them as a single baseline file.
[0159] 4. MSI status of 103 amplicon sequencing clinical validation samples was detected using MSI baseline analysis.
[0160] a) For each site s of each sample to be tested, perform the following steps:
[0161] ① Following the method described in step 2.a, extract reads from the preprocessed bam of each sample that fully cover the region where the site is located and at least two bases long on each of the two wings;
[0162] ② Following the method described in step 4.a, calculate the deletion ratio and reference ratio of the sample to be tested at site s based on the extracted reads;
[0163] ③ Following the method described in step 4.b, calculate the depth of the sample to be tested at site s based on the extracted reads, and match a suitable baseline for the sample according to the depth;
[0164] ④ Using the formula in step 4.c, calculate the Z_del of the test sample at site s based on the deletion ratio and reference ratio of the test sample and the matched baseline. s and Z_ref s ;
[0165] ⑤ Set d_cutoff to 100 and Z_cutoff to 4, and determine the state of the test sample site s according to the rules in step 4.d.
[0166] b) For each sample, count the number of stable and unstable sites at the 68 MSI sites and calculate the MSI score. Set loci_cutoff to 40, score_cutoff to 0.15, and c_cutoff to 6, and perform the first MSI status determination on the samples according to the rules in step 4.e. Based on the first determination results, 80 samples were determined to be MSS / MSI-L, and 23 samples were determined to be MSI-H.
[0167] c) Calculate ρ for the 23 samples detected as MSI-H according to the method described in step 5. Set ρ_cutoff to 0.95, and ρ for 4 samples is greater than or equal to 0.95.
[0168] d) Following the method described in step 6, adjust the baseline, re-evaluate the status of each site in the four anomalous samples, and re-count the number of stable and unstable sites, calculating the MSI score. Set loci_cutoff to 40, score_cutoff to 0.15, and c_cutoff to 6. Based on the results, the final MSI status of the four anomalous samples is determined to be MSS / MSI-L.
[0169] e) Statistical analysis of MSI detection results in clinical validation samples: In this embodiment, the sensitivity was 100% in 23 amplicon sequencing MSI-H clinical validation samples, and the specificity was 100% in 80 amplicon sequencing MSS / MSI-L clinical validation samples. Sensitivity refers to the proportion of genuine MSI-H samples that were correctly detected; specificity refers to the proportion of genuine MSS / MSI-L samples that were correctly detected. Detailed MSI scores and ρ values are shown in Table 7.
[0170] Table 7. Results of Amplicon Sequencing Clinical Validation Samples
[0171] 5. Reconstruct libraries for amplicon sequencing abnormal samples and compare the consistency of MSI results between the two library reconstructions.
[0172] For four amplicon sequencing clinical validation samples with a p-value greater than or equal to 0.95, library construction and amplicon capture were re-performed, and the reconstructed library and sequencing data were re-tested using the same method as the first MSI detection. After reconstructed library construction and sequencing, the MSI scores calculated at 68 MSI sites for all four samples were below 0.15, and the total number of stable and unstable sites was greater than or equal to 40. The results of this test were all MSS / MSI-L. The MSI detection results of the sequencing data from the two library construction experiments for the four samples are summarized (Table 8). Although the samples from the first library construction showed abnormalities, after calculating p-values and adjusting the baseline, these samples were ultimately determined to be MSS / MSI-L, consistent with the results after reconstructed library construction. This result demonstrates that the present invention can obtain relatively stable and accurate detection results even under experimental or sequencing abnormalities.
[0173] Table 8. Comparison of detection results from two library construction experiments for amplicon sequencing abnormal samples
[0174] For any part of this invention not described in detail, please refer to the prior art or the art known to those skilled in the art.
[0175] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of the present invention without departing from the spirit and scope of the claims, and all of these forms are within the protection scope of the present invention.
Claims
1. A method for detecting microsatellite instability based on next-generation sequencing technology, characterized in that, The method includes: Select a set of MSI sites with repeat sequence lengths between 10-15 bp, and design capture probes and / or amplicon primers based on any combination of the MSI sites. Select several samples, and for each sample, calculate the proportion of deleted fragments and the proportion of normal fragments at each MSI site; Several samples that were determined by PCR to be in MSI-L or MSS status were selected from the above samples as baseline samples. Different depth levels were set for each baseline sample, and a baseline was constructed based on each depth level. Based on the established baseline, MSI status detection is performed on each selected sample; Select samples with an MSI status detection result of MSI-H, and determine whether they are anomalous samples by calculating the Spearman rank correlation coefficient; if they are determined to be anomalous samples, adjust the baseline of the sample and re-determine the MSI status.
2. The microsatellite instability detection method based on next-generation sequencing technology according to claim 1, characterized in that, The calculation of the proportion of missing segments and the proportion of normal segments in a sample includes: a) Align the sequenced base sequence to the human genome reference sequence and extract sequencing base sequences that completely cover the region where the MSI site is located and at least two bases on each side; b) Filter the sequencing data of the captured sequencing samples; c) Remove duplicate sequences from the filtered sequencing base sequences of the captured sequencing samples; d) For each remaining sequencing base sequence, count the repeat sequence length at the current MSI site. Based on the relative relationship between the repeat sequence length of the sequencing base sequence and the human genome reference sequence, classify the sequencing base sequences. Count the number of sequencing base sequences in each category and the total number of sequencing base sequences, and calculate the proportion of deleted fragments and the proportion of normal fragments.
3. The microsatellite instability detection method based on next-generation sequencing technology according to claim 2, characterized in that, The filtering of sequencing data from captured sequencing samples includes: Extract three bases from each sequencing base sequence on each side of the MSI site; for MSI sites where the edges of the reads cover less than three bases, extract only two bases; if the sequencing base sequence at these positions is not completely consistent with the human genome reference sequence, filter out that sequencing base sequence. Extract the sequence corresponding to the MSI site region from each sequencing base sequence; if there are two or more bases in the sequence that are inconsistent with the reference sequence, filter the sequencing base sequence; if there is one base in the sequence that is inconsistent with the human genome reference sequence and the sequence happens to be at the start or end position of the MSI site, filter out the sequencing base sequence.
4. The microsatellite instability detection method based on next-generation sequencing technology according to claim 2, characterized in that, The process of deduplicating the sequenced bases after filtering the captured sequencing samples includes: For the captured and sequenced samples, the origin of the original DNA molecular fragment of the sequenced base sequence is determined according to the start and end sites of each sequenced base sequence; the sequenced base sequence is either a single-end sequenced base sequence or a paired-end sequenced base sequence; For repeated sequencing base sequences originating from the same original DNA molecule fragment, one base sequence is randomly selected from the repeated sequencing base sequences and retained.
5. The microsatellite instability detection method based on next-generation sequencing technology according to claim 1, characterized in that, The process of establishing a baseline includes: a) Calculate the proportion of missing fragments and the proportion of normal fragments for each baseline sample at the corresponding depth level of each MSI site; b) Calculate the baseline for each MSI site at its depth level; this includes: Calculate the baseline mean and standard deviation of the proportion of missing fragments in the baseline sample for each MSI locus; Calculate the baseline mean and standard deviation of the proportion of normal fragments in the baseline sample for each MSI locus.
6. The microsatellite instability detection method based on next-generation sequencing technology according to claim 1, characterized in that, Based on the established baseline, the MSI state detection for each selected sample includes: a) Calculate the proportion of missing fragments and the proportion of normal fragments at each MSI site for each sample; b) Calculate the depth of each MSI site and match a baseline of equal depth level based on that depth; c) Calculate the Z-score of the proportion of missing fragments and the proportion of normal fragments at each MSI site relative to the baseline; d) Determine the status of the sample at each MSI site based on the calculated depth and Z_score; The determination result is: site quality control unqualified, site unstable, or site stable; e) Based on the overall status of all MSI sites, calculate the MSI score and count the number of sites that pass quality control. Determine the MSI status of the sample based on the statistical results.
7. The microsatellite instability detection method based on next-generation sequencing technology according to claim 1, characterized in that, The Spearman rank correlation coefficient is the Spearman rank correlation coefficient between the proportion of missing fragments and the mean of the baseline proportion of missing fragments. Set an outlier threshold for the Spearman rank correlation coefficient. When the Spearman rank correlation coefficient is greater than or equal to the outlier threshold, the corresponding sample is judged as an outlier sample; otherwise, the corresponding sample is judged as normal.
8. The microsatellite instability detection method based on next-generation sequencing technology according to claim 1, characterized in that, The number of MSI sites with repeat sequence lengths between 10 and 15 bp was 74.
9. The microsatellite instability detection method based on next-generation sequencing technology according to claim 1, characterized in that, The process of adjusting the baseline of abnormal samples and re-evaluating the MSI status includes: adjusting the baseline matched at each site of the sample, re-evaluating the site status based on the adjusted baseline, combining the evaluation results of all sites to obtain the new MSI status of the sample, and replacing the original MSI status with the new MSI status.
10. A microsatellite instability detection device based on next-generation sequencing technology, comprising a processor and a memory, wherein the memory stores a software program, characterized in that, When the processor runs the software program, it can implement the microsatellite instability detection method based on second-generation sequencing technology as described in any one of claims 1 to 9.