DREAMM

DREAMM uses an ensemble machine learning algorithm to identify protein-membrane interfaces and predict nearby small-molecule binding pockets in peripheral membrane proteins (PMPs) to enable targeting of membrane-associated sites for drug design and allosteric modulation.

Key Features:

• Machine Learning-Driven Interface Identification: Employs a robust ensemble machine learning algorithm to identify protein-membrane interfaces in peripheral membrane proteins (PMPs).
• Binding Pocket Prediction: Predicts binding pockets proximal to membrane-penetrating amino acids that are accessible to small molecules.
• Conformational Ensemble Support: Operates on protein conformational ensembles provided by users or generated internally to capture dynamic pocket accessibility.
• Allosteric Modulation Targeting: Identifies interface sites suitable for allosteric modulation by drug-like molecules.

Scientific Applications:

• Targeting Peripheral Membrane Proteins (PMPs): Identification of membrane-associated and interfacial sites for small-molecule targeting and allosteric modulation of PMPs.
• Drug Design and Hit Prioritization: Prioritization of binding pockets near membrane interfaces to inform small-molecule design and screening efforts.
• Expanding Druggable Space: Enables exploration of protein-membrane interfaces as potential druggable sites for proteins that are challenging to target.

Methodology:

Interface identification using an ensemble machine learning algorithm; prediction of binding pockets near membrane-penetrating amino acids based on protein conformational ensembles provided by users or generated internally.