MutSignatures: An R Package for Extraction and Analysis of Cancer Mutational Signatures

Abstract

Cancer cells accumulate somatic mutations as result of DNA damage and inaccurate repair mechanisms. Different genetic instability processes result in distinct non-random patterns of DNA mutations, also known as mutational signatures. We developed mutSignatures, an integrated R-based computational framework aimed at deciphering DNA mutational signatures. Our software provides advanced functions for importing DNA variants, computing mutation types, and extracting mutational signatures via non-negative matrix factorization. We applied mutSignatures to analyze somatic mutations found in smoking-related cancer datasets. We characterized mutational signatures that were consistent with those reported before in independent investigations. Our work demonstrates that selected mutational signatures correlated with specific clinical and molecular features across different cancer types, and revealed complementarity of specific mutational patterns that has not previously been identified. In conclusion, we propose mutSignatures as a powerful open-source tool for detecting the molecular determinants of cancer and gathering insights into cancer biology and treatment. Keywordsmutational signatures cancer next-generation sequencing genetic instability dimensionality reduction DF, JJM designed the research project. DF, YY acquired and prepared the data. DF developed software, wrote R extension, performed data analyses. DF and VV contributed to data visualization.

Integrated genomic analysis reveals mutated ELF3 as a potential gallbladder cancer vaccine candidate

“…We detected a set of six common signatures using Non-Negative Matrix Factorization across the combined data set. Using the mutSignatures ( https://cancer.sanger.ac.uk/cosmic/signatures_v2 ) package 64 we compared our signatures to that reported in COSMIC 10 version 2. We also repeated the analysis after removing the MSI samples (Supplementary Fig.…”

Characterization of two‐dimensional materials from Raman spectral data

“…We made use of R software for the mining of literature available in the open‐access resources both manually and automated through the use of applied programming interfaces. The packages used for initial keyword‐based literature mining are rPlos , rcrossref , aRxiv , easyPubMed , NLP , and fulltext . We have also made use of AI tools like Dimensions.ai and Semantic Scholarfor doing thematic literature mining.…”