cancereffectsizeR

cancereffectsizeR quantifies the effects of somatic nucleotide mutations in cancer by estimating variant-specific selection and proliferative effect sizes.

Key Features:

• Organizing somatic variant data: Organizes somatic variant data for downstream analysis.
• Mutational signature analysis: Supports mutational signature analysis to identify patterns of mutation types.
• Site-specific mutation rate calculation: Calculates site-specific mutation rates for individual genomic loci.
• Modeling selection effects: Tests models of selection to distinguish driver versus passenger mutations by estimating selection coefficients.
• Effect estimation models: Estimates effects from single nucleotides up to gene-level aggregations.
• Epistatic effect estimation: Estimates epistatic interactions between paired sets of variants.
• Custom model design and testing: Allows design and testing of custom models for effect estimation.
• Validation and predictive power: Validation on a pan-cancer dataset showed ClinVar likely pathogenic or pathogenic variants exhibit higher estimated effects than other variants.
• Application in epistasis studies: Revealed in lung adenocarcinoma that driver mutations in BRAF, EGFR, or KRAS typically reduce selection for alterations in other genes.
• Reference data support: Provides companion reference data packages for hg19 and hg38 and a reference data builder for other species or custom genome builds given genomic and transcriptomic data.

Scientific Applications:

• Oncogenic trajectory insights: Provides estimates of mutation effects to inform analyses of oncogenic trajectories.
• Prognostic and therapeutic implications: Identifies key mutations with implications for prognosis and targeted therapy prioritization.

Methodology:

Computational steps include organizing somatic variant data, mutational signature analysis, calculation of site-specific mutation rates, testing models of selection, effect estimation from single nucleotides to genes, estimation of epistatic interactions between paired variant sets, custom model design and testing, use of reference data packages for hg19/hg38 or a user-built reference from genomic and transcriptomic data, and validation using a pan-cancer dataset compared to ClinVar pathogenicity.