AE-PocketMiner Uses Attention to Simultaneously Predict Cryptic Pockets and Their Allosteric Coupling.

Area of Science:

• Biochemistry
• Computational Biology
• Drug Discovery

Background:

• Cryptic pockets are transient protein sites crucial for drug targeting.
• Identifying and understanding allosteric regulation of these pockets is challenging.
• Current methods struggle to predict cryptic pocket locations and functional relevance.

Purpose of the Study:

• To develop an AI-driven method for simultaneous prediction of cryptic pockets and their allosteric coupling.
• To enhance the identification and targeting of cryptic pockets for expanding the druggable proteome.
• To provide a framework for drug discovery, including pocket identification, prioritization, and assay design.

Main Methods:

• Introduction of attention enabled (AE-)PocketMiner, an artificial intelligence (AI) method.
• Utilizing a graph neural network with an attention mechanism.
• Predicting cryptic pocket locations and allosteric coupling from a single protein structure.

Main Results:

• AE-PocketMiner outperforms existing methods in identifying cryptic pockets.
• The method successfully recapitulates known allosteric interactions.
• Experimental validation confirmed newly predicted cryptic pockets and their allosteric control mechanisms.

Conclusions:

• AE-PocketMiner offers a powerful framework for identifying and prioritizing cryptic pockets.
• The AI tool aids in designing assays for drug discovery targeting previously inaccessible proteins.
• This approach significantly expands the potential of the druggable proteome.