DeepAlloDriver

DeepAlloDriver predicts cancer driver mutations at protein allosteric sites to identify mutations that perturb protein structure, dynamics, and energy communication relevant to tumorigenesis and drug discovery.

Key Features:

• Deep Learning Methodology: Employs deep learning techniques to predict driver mutations with reported accuracy >93% and high precision.
• Focus on Allosteric Sites: Specifically targets allosteric sites, regions that regulate protein function distal to active sites, where mutations can influence protein structure, dynamics, and energy communication.
• Comprehensive Analysis: Identifies driver mutations at allosteric sites to aid in deciphering mechanisms underlying cancer development.
• High-throughput Prediction: Enables high-throughput prediction and analysis of protein mutations.

Scientific Applications:

• Mechanistic Insights: Provides insights into how specific mutations contribute to tumorigenesis, exemplified by identifying a missense mutation in RRAS2 (Gln72 to Leu) as an allosteric driver in knock-in mice and human patients.
• Therapeutic Target Prioritization: Pinpoints critical allosteric mutations to prioritize therapeutic targets and facilitate development of allosteric drugs.

Methodology:

Leverages a deep learning-based strategy to analyze protein structural and dynamic changes induced by mutations, enabling high-throughput prediction and analysis.