A comparative study of available software for high-accuracy homology modeling: from sequence alignments to structural...

Area of Science:

• Computational biology
• Structural bioinformatics
• Protein modeling

Background:

• Evaluating protein homology modeling software is crucial for drug discovery and biological research.
• Assessing both model quality and user-friendliness is key for practical application.

Purpose of the Study:

• To objectively assess the performance of protein homology modeling packages.
• To compare the quality of results and ease of use across different software.
• To investigate the impact of sequence identity on modeling accuracy.

Main Methods:

• Examined homology-built models for therapeutically relevant proteins with sequence identities from 19% to 76%.
• Developed novel metrics: difference alignment index (DAI) for local alignments and relative sequence alignment (RSA) for global comparisons.
• Compared sequence alignments and 3D models against structure-based alignments and crystal structures.

Main Results:

• At sequence identities >40%, all tested packages yielded similar, satisfactory results.
• For low sequence identities (<25%), Profit and Prime, incorporating structural information, showed superior sequence alignments.
• Prime generated the best model in the low sequence identity region (<25%).
• DSModeler and MOE provided reasonable models for sequence identities >25% and were significantly more user-friendly.

Conclusions:

• Software performance in protein homology modeling varies with sequence identity.
• Packages incorporating structural information, like Prime and Profit, are advantageous for low sequence identity modeling.
• DSModeler and MOE offer a practical balance of usability and model quality for higher sequence identities.