SWPepNovo

SWPepNovo performs de novo peptide sequencing from large-scale tandem mass spectrometry (MS/MS) spectra to enable high-throughput proteomic analysis.

Key Features:

• Parallel Peptide Spectrum Matches (PSMs) Algorithm: Employs a parallel PSMs algorithm with two-level parallelization using MPI-based task-level parallelism across Multiple Processing Elements (MPEs) with data transformation and dynamic feedback task scheduling, and asynchronous multithreading across Core Processing Elements (CPEs) for thread-level parallelism.
• Vectorization: Applies vectorization to process multiple data points simultaneously and reduce computational time.
• Double Buffering: Uses double buffering to overlap computation and data transfer and minimize idle time.
• Memory Access Optimizations: Implements memory access optimizations to improve memory bandwidth usage and reduce latency, addressing compute-bound and memory-bound bottlenecks.
• Scalability and Comparative Performance: Scales to datasets up to 11.22 GB and demonstrates superior efficiency and speed compared with PepNovo+, PEAKS, and DeepNovo-DIA.

Scientific Applications:

• High-throughput de novo peptide sequencing: Enables large-scale peptide identification and sequencing from MS/MS spectra for proteomic studies.
• Large-scale proteome analysis and benchmarking: Supports analysis of extensive proteomic datasets and comparative performance evaluation against other de novo sequencing tools.

Methodology:

Optimized for the SW26010 many-core processor architecture and implements MPI-based task distribution, data transformation, dynamic feedback task scheduling, asynchronous task transfer and multithreading, vectorization, double buffering, and memory access optimizations.