CanDriS

CanDriS identifies and profiles cancer-driving sites from large somatic mutation datasets using statistical models to distinguish driver from passenger mutations for pan-cancer and tumor-type analyses.

Key Features:

• Two-Component Mixture Model: Employs a two-component mixture model with a ground component representing passenger mutations and a rapidly evolving component representing driver mutations.
• Empirical Bayesian Procedure: Uses an empirical Bayesian approach to calculate the posterior probability that a specific site is cancer-driven.
• Pan-cancer and Tumor-type Profiling: Profiles potential cancer-driving sites at both pan-cancer and tumor-type levels.
• High-Confidence Sites: Identifies approximately 1% of sites with posterior probabilities greater than 0.90 as high-confidence candidate driver sites.
• Data Sources: Analyzes somatic mutation datasets from The Cancer Genome Atlas (TCGA PanCanAtlas) and the International Cancer Genome Consortium (ICGC Release 25).
• Database Integration: Integrates results into the CandrisDB repository (http://biopharm.zju.edu.cn/candrisdb/).

Scientific Applications:

• Driver-site identification: Maps candidate cancer-driving sites to support studies of somatic-cell evolution in carcinogenesis.
• Comparative mutation landscapes: Enables comparison of mutation patterns across cancers and tumor types at site resolution.
• Support for precision oncology: Provides candidate driver sites that can inform targeted therapeutic research and precision-medicine studies.

Methodology:

Analyzes somatic mutation data from TCGA PanCanAtlas and ICGC Release 25, fits a two-component mixture model separating passenger (ground) and rapidly evolving (driver) components, applies an empirical Bayesian procedure to compute site-level posterior probabilities, and profiles results at pan-cancer and tumor-type levels.