BATCAVE

BATCAVE applies a Bayesian framework to improve detection and evaluation of low-frequency somatic mutations in tumors by using tumor- and site-specific prior probabilities.

Key Features:

• Context-Aware Mutation Evaluation: Uses a Bayesian framework that incorporates tumor- and site-specific prior probabilities of mutation to adjust variant calls.
• Tumor Mutational Profile Learning: Learns each tumor's mutational profile and mutation rate from data to inform priors.
• MuTect and R implementation: Implemented in R and designed to integrate with the variant caller MuTect.
• Improved Variant Detection: In simulations, integration with MuTect improves calibration of posterior probabilities and balances precision and recall for variant detection.
• Real Data Performance: Demonstrated robust performance on real tumor sequencing data.
• Extensibility: Adaptable to other variant callers and extensible to incorporate additional biological features that influence mutation generation.

Scientific Applications:

• Low-frequency somatic mutation detection: Improves identification of somatic mutations present in subclonal tumor populations.
• Tumor heterogeneity and evolution: Enables analysis of tumor heterogeneity and evolutionary trajectories by more accurately resolving subclonal variants.
• Treatment resistance and prognosis: Supports studies of treatment resistance and patient prognosis through refined somatic mutation calls.

Methodology:

Uses a Bayesian statistical approach that learns tumor-specific mutational profiles and mutation rates and applies tumor- and site-specific priors to compute calibrated posterior probabilities for candidate somatic variants.