# Network-based coverage of mutational profiles reveals cancer genes

**Authors:** Borislav H. Hristov, Mona Singh  
**arXiv:** [1704.08544](https://arxiv.org/abs/1704.08544)  
**Date:** 2017-04-26  
**Summary page:** https://www.summarizepaper.com/en/arxiv-id/1704.08544v1/

## Key points

- Identifying specific genetic alterations in cancer genomics is a main objective
- Extensive heterogeneity observed in mutational profiles across cancer genomes
- Individual genomes exhibit numerous alterations, with only a few genes commonly mutated
- Frequently and infrequently mutated genes both play important roles in cancer development
- Novel method introduced that considers individual mutational profiles within protein-protein interaction networks
- Aims to identify small connected subnetworks of genes that collectively contribute to perturbed pathways across individuals
- Objective function balances identifying a small subset of genes and covering a significant proportion of individuals
- Solved using integer linear programming and developed a fast heuristic algorithm for practical use
- Method called nCOP outperformed methods without network information and state-of-the-art network-based methods in large-scale analysis on 6,038 tumor samples from 24 different cancer types obtained from TCGA
- Combining per-individual mutational information with interaction networks uncovers functionally relevant genes, particularly those less frequently mutated
- Study contributes to understanding cancer genetics by providing an effective approach for identifying cancer genes based on their impact on biological pathways rather than solely relying on individual gene mutation frequencies.

## Summary

In the field of cancer genomics, one of the main objectives is to identify the specific genetic alterations that drive tumor initiation and progression. However, this task is particularly challenging due to the extensive heterogeneity observed in mutational profiles across cancer genomes. Each individual's genome exhibits numerous alterations, with only a few genes being commonly mutated among different individuals. Additionally, both frequently and infrequently mutated genes play important roles in cancer development. To address these challenges, researchers have introduced a novel method that takes into account individual mutational profiles within the context of protein-protein interaction networks. This approach aims to identify small connected subnetworks of genes that may not be frequently mutated individually but collectively contribute to perturbed pathways across a large fraction of individuals. The researchers devised an objective function that strikes a balance between identifying a small subset of genes and covering a significant proportion of individuals. They solved this problem using integer linear programming and also developed a fast heuristic algorithm that performs well in practice. To evaluate their method, called nCOP, the researchers conducted a large-scale analysis on 6,038 tumor samples from 24 different cancer types obtained from The Cancer Genome Atlas (TCGA). Their results demonstrated that nCOP outperformed methods that did not utilize network information as well as state-of-the-art network-based methods that aggregated mutational information across individuals. This highlights the power of combining per-individual mutational information with interaction networks to uncover functionally relevant genes in cancers, particularly those that are less frequently mutated. Overall, this study contributes to our understanding of cancer genetics by providing an effective approach for identifying cancer genes based on their impact on biological pathways rather than solely relying on individual gene mutation frequencies.