LARVA

LARVA identifies recurrent variants in noncoding genomic regions by integrating genetic variant data with noncoding functional annotations and modeling mutation counts to estimate local mutation rates and detect noncoding driver mutations in cancer.

Key Features:

• Comprehensive integration: Integrates genetic variant data with a broad array of noncoding functional annotations and noncoding functional elements to contextualize mutations.
• Overdispersion modeling: Employs a β-binomial distribution to account for substantial overdispersion in mutation counts due to heterogeneity and correlations between neighboring sites.
• Regional genomic features: Incorporates regional genomic features such as replication timing to refine local mutation rate estimates and identify mutational hotspots.
• Recurrent variant detection: Detects recurrent noncoding variants and identifies candidate noncoding drivers, including known events such as TERT promoter mutations.
• Application to large cohorts: Demonstrated analysis on a dataset of 760 whole-genome tumor sequences to identify recurrent noncoding mutations and novel regulatory sites.

Scientific Applications:

• Cancer genomics: Identification of noncoding driver mutations and mutational hotspots relevant to tumorigenesis, exemplified by TERT promoter mutations.
• Regulatory element discovery: Detection of novel regulatory sites with elevated mutation rates that are candidate noncoding drivers of disease.

Methodology:

Integrates noncoding functional annotations and regional genomic features and models mutation counts using a β-binomial distribution, incorporating replication timing to refine local mutation rate estimates.