Toward identification of common DNA repair process in mutational signatures

Area of Science:

• Genomics
• Cancer Biology
• Bioinformatics

Background:

• Mutational signatures reveal patterns of DNA mutations from mutagenic processes.
• Understanding DNA repair pathways is crucial for cancer therapy, but tools for inferring these pathways from signatures are limited.
• Mutational signatures result from DNA damage and subsequent DNA repair.

Purpose of the Study:

• To develop a novel computational method, RePrint, for modeling the DNA repair step in mutational signature generation.
• To infer shared DNA repair mechanisms between mutational signatures using a transformation-based approach.
• To enable guilt-by-association prediction of DNA repair pathways and inform cancer treatment strategies.

Main Methods:

• Modeled the DNA repair process as a transformation (RePrint) from damaged nucleotides to repair-related mutation patterns.
• Quantified similarity between mutational signatures based on their inferred DNA repair transformations.
• Clustered signatures using RePrint similarity and compared performance against traditional signature-based clustering.

Main Results:

• RePrint similarity effectively indicates shared DNA repair mechanisms.
• RePrint-based clustering significantly outperformed signature-based clustering across multiple evaluation metrics.
• Validated guilt-by-association predictions with existing literature, identifying shared repair mechanisms even with divergent mutational profiles.

Conclusions:

• RePrint is the first systematic approach to transfer DNA repair information between mutational signatures.
• This method facilitates the understanding of signatures with unknown origins and aids in developing targeted therapeutic strategies.
• The open-source implementation of RePrint enables broader application in cancer research and treatment design.