# Machine learning methods for prediction of cancer driver genes: a survey paper

**Authors:** Renan Andrades, Mariana Recamonde-Mendoza  
**arXiv:** [2109.13685](https://arxiv.org/abs/2109.13685)  
**Date:** 2021-09-28  
**Summary page:** https://www.summarizepaper.com/en/paper/2109.13685v1/

## Key points

- Identifying specific genes and mutations in cancer research is crucial for understanding the disease and developing new approaches for diagnosis and treatment.
- Accurately pinpointing driver mutations and associated genes among millions of potential somatic mutations is a significant challenge.
- Computational methods, particularly machine learning techniques, have become important in identifying genomic patterns linked to cancer drivers and developing predictive models.
- This survey paper provides a comprehensive analysis of ML-based computational approaches used to identify cancer driver mutations and genes.
- It discusses how previous solutions have explored the interactions between different types of data and ML algorithms.
- The authors highlight current analytical limitations that require further attention from the scientific community.
- The goal is to inspire new researchers to tackle open problems and advance our knowledge in discovering cancer drivers.
- Machine learning methods play a significant role in predicting cancer driver genes, leveraging genomics data to enhance our understanding of cancer biology.

## Summary

In the field of cancer research, identifying the specific genes and mutations that contribute to the development of tumors is crucial for advancing our understanding of the disease and finding new approaches for diagnosis and treatment. However, with the vast amount of genomics data available, accurately pinpointing driver mutations and their associated genes, known as cancer driver genes, among millions of potential somatic mutations remains a significant challenge. To address this issue, computational methods have become increasingly important in identifying genomic patterns linked to cancer drivers and developing predictive models for driver events. Machine learning (ML) techniques have played a pivotal role in these efforts, offering valuable opportunities to bridge existing gaps in the field. This survey paper aims to provide a comprehensive analysis of ML-based computational approaches used to identify cancer driver mutations and genes. It offers an integrated and panoramic view of the broad data and algorithmic landscape within this scientific problem by discussing how previous solutions have explored the interactions between different types of data and ML algorithms. The authors also highlight current analytical limitations that require further attention from the scientific community. The ultimate goal is to familiarize readers with significant developments brought about by ML in this field, inspiring new researchers to tackle open problems and advance our knowledge in discovering cancer drivers. Overall, this survey paper provides an extensive overview of machine learning methods employed in predicting cancer driver genes. It emphasizes the importance of computational approaches in leveraging genomics data to enhance our understanding of cancer biology and facilitate advancements in diagnosis and treatment strategies.