DawnRank: discovering personalized driver genes in cancer

Abstract

Large-scale cancer genomic studies have revealed that the genetic heterogeneity of the same type of cancer is greater than previously thought. A key question in cancer genomics is the identification of driver genes. Although existing methods have identified many common drivers, it remains challenging to predict personalized drivers to assess rare and even patient-specific mutations. We developed a new algorithm called DawnRank to directly prioritize altered genes on a single patient level. Applications to TCGA datasets demonstrated the effectiveness of our method. We believe DawnRank complements existing driver identification methods and will help us discover personalized causal mutations that would otherwise be obscured by tumor heterogeneity. Source code can be accessed at

DriverFinder: A Gene Length-Based Network Method to Identify Cancer Driver Genes

“…The longest gene across human genome is TTN and it has been proven that higher mutation rate in it is likely to be artifacts [23,36]. For example, in BRCA, TTN ranked 4 and 6 in frequency-based method and in MUFFINN, respectively, due to high mutation rate.…”

“…In addition, iMCMC was a networkbased method by integrating somatic mutation, CNVs, and gene expressions without any prior information [6]. Another method, DawnRank, was also a network-based algorithm to discover personalized causal driver mutations by ranking mutated genes according to their potential to be drivers based on PageRank algorithm [23]. Bashashati et al developed a method called DriverNet which comprehensively analyzed genomes and transcriptomes datasets to identify likely driver genes in population-level by virtue of their effect on mRNA expression networks and also reveal the infrequent but important genes and patterns of pathway [10].…”

Integrated RNA and DNA sequencing reveals early drivers of metastatic breast cancer

“…DawnRank integrates DNA alterations, protein-protein interaction networks, and the expression of these networks via RNA gene expression data for each individual tumor. By evaluating the perturbation of the network through RNA gene expression data for each tumor, DNA alterations can be ranked in terms of the RNA networks' expression in that tumor, and thus those DNA alterations with the greatest effects in terms of RNA network expression can be identified as genetic drivers in an individual tumor specimen (14).…”

“…Many mutations and copy number alterations (CNAs) are probably passenger alterations without functional biologic consequences. We therefore used a computational tool called DawnRank (14) to identify genetic drivers. DawnRank integrates DNA alterations, protein-protein interaction networks, and the expression of these networks via RNA gene expression data for each individual tumor.…”

“…Because copy number variations (CNVs) cause many genes to be altered at once, potential genetic drivers from CNVs are often difficult to identify. Computational approaches integrating known protein-protein interaction networks, with gene expression from RNA-sequencing (RNA-seq) data and DNA-based somatic alterations have demonstrated the power of this approach for identifying unique driver sets beyond somatic mutations alone (14)(15)(16). This integrative approach could be used to identify shared drivers of breast cancer metastasis across patients and subtypes.…”