# Prediction of cancer driver genes and mutations: the potential of integrative computational frameworks

**Authors:** Mona Nourbakhsh, Kristine Degn, Astrid Saksager, Matteo Tiberti, Elena Papaleo  
**arXiv:** [2303.17402](https://arxiv.org/abs/2303.17402)  
**Date:** 2023-03-30  
**Summary page:** https://www.summarizepaper.com/en/arxiv-id/2303.17402v1/

## Key points

- The field of cancer research has been revolutionized by the vast amount of sequencing data available
- Predictive tools have been proposed to compare and prioritize driver genes and mutations, but there is little consensus on the computational approach or a golden standard for comparison
- Benchmarking these tools depends highly on input data, indicating that overfitting remains a significant problem
- One solution is to limit the scope and usage of specific tools, but this forces researchers to walk a tightrope between creating high-quality tools for specific purposes and describing the complex alterations driving cancer
- Many bioinformatic pipelines rely on single nucleotide variants or alterations in isolation without accounting for cellular compartmentalization, mutational burden, or disease progression
- Silo-research risks overlooking potential synergies or breakthroughs within bioinformatics and computational cancer biology fields
- An overview of databases and datasets for building or testing predictive tools for discovering cancer drivers is provided, including introducing predictive tools for driver genes, driver mutations, and their impact based on structural analysis
- The authors suggest moving towards integrative frameworks to avoid silo-research that consider multiple factors such as cellular compartmentalization and disease progression when analyzing genetic alterations in cancer cells
- By doing so, researchers can develop more accurate predictive models that consider the complexity of cancer biology
- Collaboration across different research fields is needed to advance our understanding of cancer genetics and develop effective treatments.

## Summary

The field of cancer research has been revolutionized by the vast amount of sequencing data available, allowing scientists to explore a range of genetic variables that may drive and progress cancer. Predictive tools have been proposed to compare and prioritize driver genes and mutations, but there is little consensus on the computational approach or a golden standard for comparison. Benchmarking these tools depends highly on input data, indicating that overfitting remains a significant problem. One solution is to limit the scope and usage of specific tools, but this forces researchers to walk a tightrope between creating high-quality tools for specific purposes and describing the complex alterations driving cancer. While knowledge about cancer development increases every day, many bioinformatic pipelines rely on single nucleotide variants or alterations in isolation without accounting for cellular compartmentalization, mutational burden, or disease progression. Even within bioinformatics and computational cancer biology fields, silo-research risks overlooking potential synergies or breakthroughs. To address this issue, an overview of databases and datasets for building or testing predictive tools for discovering cancer drivers is provided. This includes introducing predictive tools for driver genes, driver mutations, and their impact based on structural analysis. The authors suggest moving towards integrative frameworks to avoid silo-research. These frameworks would consider multiple factors such as cellular compartmentalization and disease progression when analyzing genetic alterations in cancer cells. By doing so, researchers can develop more accurate predictive models that consider the complexity of cancer biology. Overall, this paper highlights the need for collaboration across different research fields to advance our understanding of cancer genetics and develop effective treatments.