GOSSIP: a method for fast and accurate global alignment of protein structures

Area of Science:

• Structural bioinformatics
• Computational biology
• Biochemistry

Background:

• The Protein Data Bank (PDB) is rapidly expanding, necessitating efficient methods for analyzing vast structural data.
• Existing protein structure comparison tools often focus on local alignments, which are computationally intensive and slow for large-scale analysis.
• There is a growing need for fast tools to identify similar structures and cluster them based on structural resemblance.

Purpose of the Study:

• To develop a novel computational method for fast, global, all-against-all alignment of protein structures.
• To enable efficient analysis of large datasets of protein structures by identifying similarities at a high speed.
• To provide a tool that overcomes the limitations of local alignment methods for large-scale structural comparisons.

Main Methods:

• Introduction of GOSSIP (Global Structure Similarity Identification Program), a novel method for global all-against-all protein structure alignment.
• GOSSIP detects structural similarities down to a user-defined cutoff, facilitating rapid identification of related structures.
• The method was evaluated on datasets of short structural fragments and large, sequence-diverse structural benchmarks.

Main Results:

• GOSSIP achieves significantly higher speeds compared to existing local structure alignment methods and database scanning approaches.
• The method demonstrates speed improvements of several orders of magnitude for comparing numerous structures.
• Evaluations confirmed that for similarity thresholds of 0.6 and above, GOSSIP maintains accuracy in alignments and the number of detected global similarities.

Conclusions:

• GOSSIP provides a computationally efficient solution for analyzing large protein structure datasets.
• The method enables high-speed identification and clustering of structurally similar proteins without compromising accuracy.
• GOSSIP is available as a server and downloadable executable, facilitating its use in structural bioinformatics research.