ZH Sem Bioinfo - Qingyao Huang

Discovering copy number variation across multiple cancer types


Genomic variations are direct cause of tumor formation and accomplice in its continuous evolution. While point mutations can be pinpointed to a targeted genetic element, copy number variations (CNVs) involve copy number gain or loss of a large DNA segment which often covers hundreds of genetic elements in one event. Although the vast majority of variations are not directly fucntionally cancer promoting, we observe consistency in CNV landscape within the same cancer types and corresponding increase in heterogeneity along with increased distinction in physiology and morphology. This implies that particular CNV may promote cancer type-specific progression. Previously, it was observed that a focal CNV (with length shorter than 3Mb) indicated stronger functional relevance than large-scale CNV. In addition, we observe higher frequency of CNV segment start and end points near known driver genes. Taken together, we designed a statistic that harness properties of collective CNV segments, to delineate the non-random alteration of genes exerted through CNV. Using 9 different cancer types from multiple sources and platforms, we ranked all genes in the genome for their importance based on the CNV data. The same cancer types cluster together by genome-wide significance scores. Known oncogenes and tumor suppressor genes can be found at the top. We have also compared with GISTIC2.0 using the same data and identified similar regional peaks, but this method can provide significance measure on the gene level. With the confirmatory results on the known tumor promoting genes, this work has a potential to identify novel functional pathways that are exerted through CNV.


Edit on Github...