SW-Tandem

SW-Tandem performs large-scale peptide identification from tandem mass spectrometry (MS/MS) data for shotgun proteomics using a parallelized spectrum dot product scoring algorithm.

Key Features:

• Parallelized scoring: Implements a parallelized spectrum dot product scoring algorithm for peptide-spectrum matching.
• High-performance target: Targets the Sunway TaihuLight supercomputer and its SW26010 many-core processors, achieving peak computation performance exceeding 100 petaflops.
• Low-level optimizations: Employs memory-access optimizations, double buffering, and vectorization to maximize computational efficiency on the SW26010 architecture.
• Scalability: Demonstrated scalable processing for datasets up to 12 GB.
• Benchmark performance: Outperformed X!! Tandem, MR-Tandem, and MSFragger in reported experimental evaluations.
• Implementation: Implemented in C++.

Scientific Applications:

• Shotgun proteomics peptide identification: Identification of peptides from tandem mass spectrometry data in shotgun proteomics workflows.
• Large-scale high-throughput proteomics: Processing of large MS/MS datasets for high-throughput proteomics studies requiring rapid peptide identification.

Methodology:

Uses a parallelized spectrum dot product scoring algorithm implemented in C++ and optimized for Sunway TaihuLight's SW26010 many-core processors via memory-access optimizations, double buffering, and vectorization, with demonstrated peak performance exceeding 100 PFlops and scalability to 12 GB datasets.