Accurate Detection Of Low-frequency Mutations Based On DNBSEQ High-throughput Sequencing Technology

Background:DNBSEQ high-throughput sequencing technology is widely used in scientific research and clinical fields,including whole-genome sequencing(WGS),whole-exome sequencing(WES)and some special targeted region sequencing(panel sequencing).With the rapid development of high-throughput sequencing technology and its application in research of cancer and other diseases,the demand for precise detection of mutations,especially for low-frequency mutations,is becoming higher and higher.However,due to the limitations of sample type,sequencing depth and other conditions,how to filter false positive mutations and accurately identify the true low-frequency mutations in samples is still a big challenge.So,it is particularly important to improve the detection accuracy of low-frequency mutations through some effective methods.Purpose:In this paper,we mainly explore and study the methods that can improve the accuracy of low-frequency mutation detection under the DNBSEQ high-throughput sequencing technology.This includes unique dual barcode(UDB),unique molecular identifier(UMI),and sequencing depth.Methods:First of all,32 DNA fragments with known sequences were prepared by single barcode and unique dual barcode(UDB)libraries and pooled for sequencing,and the contamination rate(index hopping)of the two libraries was compared.Second,UMI was used to detect the variation of circulating tumor DNA(ctDNA)with expected variant allele frequency(expected VAF)of 1%,0.5%and 0.2%,as well as ctDNA samples from 8 cervical cancer patients.The effect of UMI on the detection of true low frequency mutations was analyzed.In the end,NA12878 with expected VAF of 3%,ctDNA standard samples with expected VAF of 1%,3%,and 5%,respectively,and the DNA from fresh frozen tissues(FF)and formalin-fixed paraffin-embedded tissues(FFPE)of C57BL/6 mice were sequenced at high depth.Then,we analyzed the effects of different sequencing depths on the fidelity and sensitivity of low-frequency mutation detection,and the establishment of a standard model of sequencing depth was studied.Results:1.UDB has more advantages in solving sample crosstalk than single barcode.The crosstalk between single barcode libraries(contamination rate)is 0.058%,while the contamination rate of UDB library is significantly reduced to as low as 0.0063%,which greatly reduces the interference to low-frequency mutation detection.2.When using UMI for mutation detection and analysis of ctDNA standards with theoretical expected VAF of 1%,0.5%and 0.2%,the interference of false positive mutation background was removed and the accuracy of detection was improved.The detection rates of real mutations were 100%(7/7),85.7%(6/7)and 83.3%(5/6),respectively.The changes in the number of ctDNA mutations before and after cervical cancer and the consistency of ctDNA and tissue mutation detection proved the clinical application value of UMI technology.3.Based on binomial distribution,the normalized relationship of different low frequency mutation-sequencing depth-sensitivity was established,and the relationship between sequencing depth and the fidelity of detected VAF was clarified.It was found that the detected VAF fluctuated regularly around the expected VAF in NA12878 with expected VAF of 3%and ctDNA with expected VAF of 1%,3%and 5%.The higher the sequencing depth is,the fluctuation of detected VAF decreased with the increase of sequencing depth,which was closer to the expected VAF,that is,the higher the fidelity,thus improving the detection sensitivity.Based on the above results and rules,when the depth of the real samples of FF and FFPE mice was also increased,the mutation consistency of both mice was also significantly improved.Conclusion:1.UDB can solve the crosstalk problem of samples and lay a foundation for accurate detection of low frequency mutation.2.The use of UMI can distinguish true mutations from false positive mutations in samples and improve the accuracy of detection.3.The improvement of sequencing depth increases the fidelity of low-frequency mutation frequency and detection sensitivity,while the standardization of sequencing depth contributes to the accurate detection of low-frequency mutation.