Toward high-throughput, multicriteria protein-structure comparison and analysis.

Area of Science:

• Biomedical research
• Structural bioinformatics
• Computational biology

Background:

• Protein-structure comparison (PSC) is crucial for drug design, protein folding, and structure prediction.
• The multicriteria PSC (MC-PSC) problem requires diverse similarity notions for various applications.
• Current MC-PSC methods are time-consuming for large-scale datasets.

Purpose of the Study:

• To develop a high-throughput distributed reimplementation of ProCKSI for MC-PSC on very large datasets.
• To enable real-time or near real-time protein structure comparisons for dedicated users.
• To significantly reduce the time required for large-scale protein structure comparisons.

Main Methods:

• Designed an innovative distributed algorithm for cluster/grid environments.
• Implemented multiple popular PSC methods (USM, MaxCMO, FAST, DaliLite, CE, TMAlign).
• Incorporated a distributed consensus-building procedure for similarity assessment.

Main Results:

• Achieved ProCKSI's similarity assessment quality with a fraction of the original time.
• Demonstrated efficient target-against-all comparisons for large protein structure datasets.
• Showcased efficient all-against-all comparisons for very large-scale datasets.

Conclusions:

• The proposed distributed method significantly accelerates MC-PSC for large datasets.
• This advancement is key for enabling real-time structural bioinformatics applications.
• Further challenges remain in achieving true real-time MC-PSC.