AlphaFold2

AlphaFold2 predicts protein three-dimensional structures from amino acid sequences with high accuracy using artificial intelligence.

Key Features:

• Protein Structure Prediction: Predicts protein structures from nearly all known protein sequences with state-of-the-art accuracy.
• Multiple Sequence Alignment Processing: Processes multiple sequence alignments as input for structure prediction, with targeted column masking enabling conformational sampling.
• Conformational Landscape Sampling: Partially samples conformational landscapes and estimates population ratios of alternative conformations through targeted column masking of multiple sequence alignments.
• Fragment-Based Prediction: Supports fragment-based approaches for predicting interactions involving short linear motifs in proteins that exceed computational feasibility at full length.
• Protein-Protein Interaction Prediction: Predicts structural models for protein-protein interactions mediated by short linear motifs, including LC3-interacting regions and SUMO-SIM interactions.

Scientific Applications:

• Drug Discovery: Provides protein structural models for virtual drug screening and molecular docking to identify ligand binding poses and competitive inhibitors, particularly for G protein-coupled receptors.
• Structural Biology: Generates protein structural models for comparison with experimental structures from X-ray crystallography and Cryo-EM.
• Autophagy Research: Predicts LC3-interacting region interactions and structural models for selective autophagy cargo recruitment mechanisms.
• Protein Dynamics Characterization: Characterizes alternative conformation populations and protein dynamics for engineered proteins, including green fluorescent protein variants with alternative β-strands.
• Protein Engineering: Supports computational design of protein-based systems for sensing applications through rapid estimation of relative conformational populations.

Methodology:

AlphaFold2 uses artificial intelligence to predict protein structures from amino acid sequences, processing multiple sequence alignments with column masking options for conformational sampling. The system employs a fragment-based approach for predicting short linear motif-mediated interactions, with fragment length and phosphomimetic mutations modulating predicted interactions.