Title:SIGH: identifying significant intratumor genomic heterogeneity in next-generation sequencing data

Abstract:Intratumor heterogeneity is both a major confounding factor in studying individual clones and an information source in characterizing tumor system. Intratumor heterogeneity cannot be resolved readily by global sequencing. Systematic efforts to characterize intratumor genomic heterogeneity must effectively distinguish significant genuine clonal sequences from probabilistic fake derivatives. We developed a software tool (SIGH) for identifying significant intratumor genomic heterogeneity directly from mixed sequencing reads that can improve genomic analyses in many oncological contexts. This open source software offers several attractive features: (1) it is supported by a well-grounded statistical framework that exploits probabilistic characteristics of sequencing errors; (2) it identifies genuine clonal sequences from raw sequence reads using significance tests; (3) its unbiased strategy allows detecting rare clone(s) despite that clone relative abundance; and (4) it estimates constituent proportions in each sample. SIGH also provides the users with the ability to utilize parallel computing on ubiquitous multicore machines.