COMER2

COMER2 performs sensitive and specific remote homology searches of protein sequences using GPU-accelerated profile–profile comparisons.

Key Features:

• GPU Acceleration: Implementation uses CUDA-enabled GPUs to accelerate computations and achieves up to 20× speedup compared to HHsearch, which relies on vectorized instructions on modern CPUs.
• Profile–Profile Comparison: The core method is the COMER profile–profile comparison that aligns sequence profiles to enhance detection of remote homologs.
• Performance Efficiency: COMER2 incorporates a rewritten COMER algorithm to improve computational efficiency while preserving sensitivity and specificity in homology detection.

Scientific Applications:

• Protein Structure Prediction: Identification of homologous proteins to inform structural modeling and comparative modeling efforts.
• Functional Annotation: Detection of distant evolutionary relationships to support annotation of newly sequenced proteins.
• Comparative Genomics and Evolutionary Studies: Analysis of protein sequence similarities to explore evolutionary links between species.

Methodology:

Profile–profile sequence alignment using the COMER method and sequence profiles, implemented with CUDA-enabled GPU acceleration and a rewritten COMER algorithm for improved performance.