OPUS-Dom: applying the folding-based method VECFOLD to determine protein domain boundaries

Area of Science:

• Computational biology
• Structural bioinformatics
• Protein structure prediction

Background:

• Accurate prediction of protein domain boundaries is crucial for understanding protein function and evolution.
• Existing methods for de novo domain prediction face challenges in accuracy and robustness.

Purpose of the Study:

• To introduce OPUS-Dom, a novel de novo method for predicting protein domain boundaries.
• To evaluate the performance of OPUS-Dom against state-of-the-art domain prediction algorithms.

Main Methods:

• OPUS-Dom utilizes VECFOLD, a coarse-grained folding method, to generate low-resolution structural models from protein sequences.
• It employs a domain parsing algorithm to identify boundaries in VECFOLD-generated decoys.
• Consensus domain boundaries are determined using statistical analysis and empirical sequence-based profiles.

Main Results:

• OPUS-Dom demonstrated superior performance compared to several existing domain prediction algorithms on benchmark datasets.
• Despite inherent errors in individual VECFOLD structures, the ensemble approach yielded robust domain boundary predictions.
• The method's success indicates that domain arrangement is primarily driven by tertiary packing of secondary structures.

Conclusions:

• OPUS-Dom offers a robust and accurate approach for de novo protein domain boundary prediction.
• The findings suggest that protein domain organization is governed by general packing principles rather than specific sequence constraints.