GalaxyDomDock

GalaxyDomDock predicts relative orientations of known protein domains using ab initio domain-domain docking to model multi-domain protein structures.

Key Features:

• Ab Initio Docking Approach: Employs an ab initio domain-domain docking algorithm based on principles from GalaxyTongDock to predict domain orientations without relying on complex templates.
• Geometric Filtering with Dijkstra Algorithm: Uses a geometric filter that applies the Dijkstra algorithm to enforce linker-derived distance and connectivity constraints and remove infeasible domain orientations.
• Loop Sampling via FALC Method: Predicts domain linker conformations by sampling loop conformations using the Flexible Anchor Loop Closure (FALC) method.
• Performance and Benchmarking: Shows superior or comparable performance to AIDA and Rosetta on benchmark sets, including two-domain proteins and discontinuous domains, and handles cases where domain structures were modeled with AlphaFold.

Scientific Applications:

• Conformational Ensemble Modeling: Predicts ensembles of domain orientations to represent alternative states of multi-domain proteins.
• Structure–Function Analysis: Supports investigation of how domain orientations and conformational variability affect protein function.
• Structure-Guided Design: Provides domain-orientation hypotheses relevant to structural biology and structure-based drug design.

Methodology:

Performs ab initio domain-domain docking based on GalaxyTongDock principles, applies a Dijkstra-based geometric filter using linker constraints, samples linker loops with the FALC method, and evaluates performance against Rosetta and AIDA benchmark sets including AlphaFold-modeled domain structures.