A New Strategy Integrating Purity And CNVs Into Somatic Mutation Calling

BackgroundSomatic mutation detection based on genome and exome sequencing data in tumor tissues is a crucially important step in tumor genomics analysis,which has been widely used in the discovery of cancer-related driver genes,the analysis of molecular evolution of tumor,the study of tumor subclone,the recognition of neoantigen,and so on.However,due to the high heterogeneity in tumor cells and diverse stromal cells and immune cells in tumor microenvironment,the variety and complexity of components in cancer tissues present huge influence on biological analysis and genome analysis.It is reported that the contamination in tumor cells affect the tumor analysis,which may result in different biological explanations.MethodFirstly,to analyze the correlation between the number of somatic mutation,tumor purity and copy number variation,we utilized 12 cancer types.Then,we proposed a new method(iPPMC)integrating purity and CNVs into consideration to call somatic mutation.In iPPMC,the proportion of mutated reads in tumor tissues(mutation rate)and reads were predicted by tumor purity and ploidy(or copy number),and the significance of difference of mutation rate between normal sample and tumor samples was detected by fisher ' s exact test.In order to verify the efficiency of iPPMC,we simulated 1000 case and matched normal samples with different sequencing depths,tumor purity and ploidy.Finally,the somatic mutations of 652 samples from TCGA were detected by iPPMC.ResultIn our study,we found that the number of mutations we detected was positively correlated with tumor purity,which suggests that the tumor purity has badly influence on somatic mutation detection.And the detection of somatic mutation is also closely related to copy number variation.It is suggestted that the tumor purity and copy number variation can lead to false-negative results in somatic mutation detection.With a large quantity of simulated data,we found that a similar number of mutated sites were detected by iPPMC for each sample with various tumor purity and CNV,which exhibited a fairly fine stability.And the recall rate of mutated sites called by iPPMC(66%)is significantly higher than Mutect(35%)and Varscan2(40%).Therefore,iPPMC can eliminate the effect of tumor purity and copy number variation on somatic mutation detection,and discover the neglected mutation sites.In the following study,the samples from TCGA detected by iPPMC show:(a)the intertumoral heterogeneity of mutation rate of LUAD,LUSC,and BRCA is not significant;(b)the somatic mutations detected by iPPMC were negatively correlated with tumor purity;(c)PTK2B(mutation rate is 0.39)and DCLK1(mutation rate is 0.36)were identified as novel BRCA-related genes with high mutation frequency,and the significant differences of survival time between mutant group and wildtype group were found(the median survival time of PTK2B_mutant and PTK2B_wildtype is respectively 2573 and 3959 days,HR=2.785,LogRank test p=0.0002;the median survival time of DCLK1_mutant and DCLK1_wildtype is respectively 2469 and 3126 days,HR=1.9674,LogRank test p=0.0012);(d)the overall mutation rates detected by iPPMC in Basal-like and HER2-enriched subtypes were the highest,while the lowest in the Normal-like subtype;(e)the mutation rate of TP53 detected by iPPMC(0.40)was higher than MuTect(0.31).ConclusionThe effect of tumor purity and copy number variation on somatic mutation calling may lead to false negative results.So it is important to improve the new method of somatic mutation detection,which considers the complexity of tumor tissue and copy number variation.iPPMC,integrating tumor purity and DNA copy number variation,reduces false negative results,which will contribute to the identification of new neoantigens and show clinical value.