A Method For Estimating Tumor Purity Using DNA Methylation Chip Data And Consensus Information Differentiation Sites

Tumor purity is an intrinsic property of tumor samples and potentially has severe impact on many types of data analysis.We have previously developed a statistical method,Infinium Purify,which could infer purity of a tumor sample given its tumor type(available in TCGA)or a set of informative CpG(iDMC)sites.However,in many clinical practices,researches may focus on a type of tumor samples that is not included in TCGA,and sample number is too few to identify reliable iDMCs.In such case,a universal set of informative CpG sites that could infer tumor purity for any given tumor type is urgently needed.In this paper,we identify a universal set of iDMCs,termed as uiDMCs,by comparing a set of elaborately selected tumor and normal samples from different tumor types in TCGA.Then for a specific tumor sample,these uiDMCs are attributed to hyper-or hypo-methylation group according to their methylation level in this tumor sample and that in normal samples.Based on these uiDMCs and their hyper-/hypo-methylation statuses,we inferred tumor purities for all samples from TCGA using a voting strategy of their transformed methylation levels.Tumor purities of 8830 tumor samples from TCGA by our method are highly consistent with those by other available methods,even from different types of data.In particular,the Spearman's correlation of our results and those by ABSOLUTE achieves 0.83,which is higher than previous result.These results suggest that tumor purity could be reliably estimated from a universal set of CpG sites.Consequently,the updated version of InfiniumPurify,named ui InfiniumPurify,can be applied to a single sample(or a few samples)of interest whose tumor type is not included in TCGA.This characteristic will greatly broaden the application of ui InfiniumPurify in cancer research.