ThreaDomEx: a unified platform for predicting continuous and discontinuous protein domains by multiple-threading and...

Area of Science:

• Computational Biology
• Structural Bioinformatics
• Protein Structure Prediction

Background:

• Accurate protein structure prediction is crucial for understanding biological function.
• Identifying continuous (CD) and discontinuous domains (DCDs) is a key challenge in structure prediction.

Purpose of the Study:

• To develop a novel hierarchical pipeline, ThreaDomEx, for predicting both CD and DCD structures.
• To improve the accuracy and scope of automated protein domain prediction.

Main Methods:

• ThreaDomEx utilizes sequence threading through the Protein Data Bank (PDB) to identify structure templates.
• A domain conservation score (DC-score) profile guides domain-segment assignment.
• Boundary-clustering algorithms refine DCD locations, and symmetry-guided assembly aids in DCD detection when templates are insufficient.

Main Results:

• ThreaDomEx achieved a 89.3% normalized domain overlap score on 1111 proteins, significantly exceeding state-of-the-art methods.
• It successfully recalled 26.7% of DCDs with 72.7% precision for proteins where other methods failed.
• The ThreaDomEx server allows interactive refinement of domain models.

Conclusions:

• ThreaDomEx offers a robust and accurate method for predicting continuous and discontinuous protein domains.
• The interactive features of the ThreaDomEx server empower users to refine predictions, especially for challenging targets.
• This tool has significant implications for structural biology research and drug discovery.