cTaG

cTaG classifies tumor suppressor genes (TSGs) and oncogenes (OGs) from somatic mutation data to identify cancer driver genes, including low-frequency drivers, across multiple cancer types.

Key Features:

• Pan-Cancer Model: Employs a pan-cancer modeling approach to analyze somatic mutation data across multiple cancer types.
• Functional Impact and Recurrence Analysis: Extracts features that capture the functional impact of mutations and their recurrence across samples to detect drivers independent of mutation frequency.
• Mutation Type Profiling: Classifies genes based on mutation-type profiles, emphasizing specific mutation types such as nonsense mutations in the classification.
• Overcoming Overfitting: Incorporates measures to avoid overfitting, improving robustness of predictions across datasets and cancer types.
• Tissue-Specific Driver Gene Identification: Can be applied to identify tissue-specific driver genes to infer context-specific functional roles.

Scientific Applications:

• Identification of Known and Novel Driver Genes: Predicts known cancer driver genes such as ARID1A, TP53, and RB1 and candidate novel drivers including CD36, ZNF750, ARHGAP35 as TSGs and TAB3 as an OG.
• Enhanced Recall and Reduced Bias: Improves recall and reduces bias toward genes with high mutation rates, facilitating discovery of low-frequency driver genes.

Methodology:

Pan-cancer modeling of somatic mutation data; feature extraction capturing functional impact and mutation recurrence; classification based on mutation-type profiles with emphasis on nonsense mutations; and strategies to prevent overfitting.