MutSpot

MutSpot identifies recurrent non-coding somatic mutation hotspots in cancer genomes by modeling position- and patient-specific background mutation probabilities using local nucleotide context, replication timing, and epigenomic features.

Key Features:

• Background Mutation Model: Constructs a background mutation model that accounts for local nucleotide context, replication timing, and epigenomic features to establish baseline mutation probabilities.
• Poisson Binomial Modeling: Uses a Poisson binomial statistical model to assess mutation recurrence in focal DNA regions while accommodating varying mutation rates across tumors.
• Feature Selection and Estimation: Performs feature selection across hundreds of epigenetic and sequence features and estimates position-specific and patient-specific background somatic mutation probabilities.
• Scalability: Scales to analyze thousands of cancer genomes and operates efficiently on standard workstations.

Scientific Applications:

• Non-coding mutation hotspot detection: Detects non-coding mutation hotspots by identifying regions with significantly higher mutation recurrence compared to background genomic mutation rates.
• Interpretation of tumorigenic mechanisms: Provides insights into oncogenic processes and regulatory disruptions by linking recurrent non-coding mutations to genomic and epigenomic context.
• Candidate target prioritization: Prioritizes genomic regions for downstream investigation as potential therapeutic or biomarker targets in cancer development.

Methodology:

Constructs a background mutation model incorporating local nucleotide context, replication timing, and epigenomic features; performs feature selection across hundreds of epigenetic and sequence features; estimates position-specific and patient-specific background somatic mutation probabilities; and applies a Poisson binomial model to assess mutation recurrence in focal DNA regions.