Machine learning predictions improve identification of real-world cancer driver mutations.

A preprint abstract from biorxiv.org discusses the use of machine learning in identifying cancer driver mutations. The study found that computational methods incorporating protein structure or functional genomic data outperformed methods trained only on evolutionary data in identifying known cancer drivers. The researchers also validated the association of pathogenic variants of KEAP1 and SMARCA4 with worse survival in patients with non-small cell lung cancer. Despite primarily training on germline mutation data, the computational predictions contributed to a more comprehensive understanding of tumor genetics. However, it is important to note that this preprint has not been peer-reviewed. [Extracted from the article]