Predicting potential cancer driver genes using hybrid approach

Abstract

Identifying cancer driver genes remains great significance since it assists in increasing the survival rate by defining cohesive treatments at early stages. Not only single algorithms but also hybrid approaches to identify driver genes do exist, but systematic ways to combine and optimize the existing algorithms on large datasets are few. By identifying the drawbacks of existing cancer driver genes identification methods, this approach formulates an effective hybrid method (Dots Witer) to identify potential cancer driver genes in cancer. The Dots Witer pipeline summarizes somatic mutations, genes involved in tumorigenesis. The input pancancer dataset consists of 2397 small somatic variants of Breast Invasive Carcinoma and 1017 small somatic variants of Lung Adenocarcinoma. Dots Witer pipeline can be applied to genes that are targeted by single nucleotide variants (SNVs) and small insertions and/or deletions (indels). The Dots Witer integrates the tools, DOTS Finder and WITER in order to identify the driver genes efficiently and effectively. This pipeline identifies 656 cancer progression genes out of 1438 genes in Breast carcinoma and 42 cancer progression genes out of 102 in Lung Adenocarcinoma. Since existing tools shows compatibility issues due to technological stack of each tool, the Dots Witer provide a consistent and common platform to execute the given exome/genome sequence dataset. Moreover due to the limitation of the processing power and the storage of the workstation, Dots Witer provides a distributed solution to scatter the ensemble approach. Compare to the existing cancer driver gene detection algorithms, this pipeline gives a higher fraction of predicted driver genes by integrating Fisher’s method.