COMER2: GPU-accelerated sensitive and specific homology searches

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Homology searching is crucial for analyzing large sequence datasets.
• Existing methods often face speed limitations when dealing with massive amounts of data.
• Sensitive homology detection is essential for biological discovery.

Purpose of the Study:

• To present a significantly improved version of the COMER homology search method.
• To develop a sensitive homology search tool capable of rapid database searching.
• To leverage GPU acceleration for enhanced performance in sequence analysis.

Main Methods:

• Developed COMER2, a rewritten homology search method based on protein sequence profile alignment.
• Implemented CUDA-enabled graphics processing unit (GPU) acceleration.
• Optimized for searching large sequence databases.

Main Results:

• COMER2 demonstrates high sensitivity in homology detection.
• The method is capable of searching large databases on standard hardware, including laptops.
• COMER2 achieves up to a 20-fold speed increase compared to HHsearch, a CPU-based state-of-the-art method.

Conclusions:

• COMER2 offers a substantial speed improvement for sensitive homology searches.
• The software is accessible, cross-platform, and open-source, facilitating wider adoption.
• This advancement enables faster and more efficient analysis of large-scale biological sequence data.