High-Performance Computing in Tandem Mass Spectrometry (MS/MS) Peptide Identification

Tandem mass spectrometry (MS/MS) has become a cornerstone technology in large-scale proteomics, producing vast datasets that present considerable computational challenges for peptide identification. Conventional database-search-based approaches often exhibit limited efficiency, particularly as protein sequence repositories expand rapidly and experimental spectra grow increasingly complex. High-performance computing (HPC) offers a viable solution by supporting large-scale parallelization and algorithmic optimization. In this study, we introduce SW-Tandem, an efficient framework for peptide identification that implements parallelization of the spectrum dot product scoring algorithm on the Sunway TaihuLight supercomputer. Through the incorporation of memory-access optimization, double buffering, and vectorization techniques, SW-Tandem achieves marked improvements in computational performance while maintaining identification accuracy. Experimental evaluations reveal that SW-Tandem substantially accelerates peptide identification and surpasses existing state-of-the-art methods when deployed across multiple nodes. These findings underscore the potential of integrating HPC with algorithmic enhancements to advance large-scale bioinformatics research.