Allelic Imbalance In Four Tumor Types

Cancer progression involves multiple genetic events,which can activate dominant-acting oncogenes and disrupt the function of specific tumor suppressor genes.Somatic DNA alterations are crucial for the acquisition of tumor-related traits.One class of alterations,allelic imbalance(AI),occurs when a segment of one parental chromosome increases or decreases in copy number relative to the other.If the homolog with the resulting larger copy number-referred to herein as the promoted homolog-carries a genetic variant that is more advantageous to tumor growth than that carried by the other homolog,then cells promoting the advantageous allele gain a selective advantage.That is,AI can be viewed as a substrate for allelic selection,facilitating identification of genes and alleles of importance.With the advent of "next generation" sequencing(NGS),it is now possible to interrogate the entire genome in an unbiased manner.Single nucleotide polymorphism(SNP)can be used to detect somatic AI events.Furthermore,the Cancer Genome Atlas(TCGA)(The Cancer Genome Atlas Research Network,2008)provides the largest collection of whole-exome sequencing(WES)data from both tumor samples and the matched normal samples.These data sets enable the potential of a near-comprehensive view of somatic AI in germline exonic polymorphisms across thousands of patient samples.The global characterization of these classes of germline coding variation now allows an agnostic and systematic search for allelic selection.Genomic studies of large tumor sample sets typically focus on recurrence as a signature of a driver/causal status.Recurrence is usually measured either with regard to point mutations and indels,or structural amplification and deletion.Here,we offer alternative and complementary approaches that combine information from variants and copy number alterations.We hypothesized that variants contributing to important tumor-related traits-whether the variant is inherited or somatically acquired-would produce an observable statistical signal of promotion in regions of AI.That is,AI is a lens through which to view selection.Specifically,we posit that common germline coding variants advantageous to tumor growth will produce a statistical signal of preferential promotion across samples.Alleles observed to be promoted across independent tumors provide evidence of cancer relevance.Here,we provide a Bayesian model to evaluate SNP-level allelic-imbalance(AI)and thereby performed exome-wide selection of AI variants in five cancer cohorts(BRCA,LUAD,PRAD,COAD)from "The Cancer Genome Atlas",we aimed to identify novel genes and variants displaying a signature of selection in tumor population,thereby implying importance for tumor related properties.We validate the model in two aspects and then perform the association analysis between AI and somatic copy number alterations.The genes being affected by allelic-imbalanced variants are enriched for relevant biological processes in the tumor,for example,AI genes of four tumors are all enriched in "KEGG PATHWAYS IN CANCERKEGG","ECM RECEPTOR INTERACTION" and "KEGG FOCAL ADHESION".We also demonstrated that AI in the DNA level is correlated with the allelic-specific transcript abundances,reconfirming the functional and phenotypical implications of the somatic selection.Our study provides a unique and complementary perspective to understand the somatic alterations in human cancer.