DPAM

DPAM parses and classifies globular domains in AlphaFold protein models to assign evolutionary context and support functional annotation of predicted structures from the AlphaFold Database (>200 million models).

Key Features:

• Automatic Domain Recognition: Identifies globular domains in AlphaFold models using inter-residue distances and predicted aligned errors (PAE) combined with ECOD domain information.
• ECOD Integration via HHsuite and Dali: Incorporates ECOD domains identified through HHsuite sequence similarity searches and Dali structural similarity searches to inform classification.
• Benchmark Performance: In a benchmark of 18,759 AlphaFold models, recognized 98.8% of domains and assigned accurate boundaries for 87.5% of domains, outperforming existing structure-based domain parsers and ECOD homology-based assignments.
• Large-scale Applicability: Designed to operate at scale on models from the AlphaFold Database to enable integration of predicted structures into evolutionary hierarchies.

Scientific Applications:

• Evolutionary Classification: Assigns domains to ECOD evolutionary hierarchies to provide evolutionary context for predicted structures.
• Functional Annotation: Delineates domain boundaries in AlphaFold models to support functional annotation of proteins.
• Structural Biology and Target Prioritization: Informs studies of protein mechanisms, interactions, and potential therapeutic target identification.

Methodology:

Analyzes inter-residue distances in 3D structures, uses predicted aligned errors to refine domain boundaries, and integrates ECOD domains obtained via HHsuite sequence searches and Dali structural searches.