MR-MSPOLYGRAPH

MR-MSPOLYGRAPH performs parallelized peptide–spectrum matching using the MapReduce framework to combine protein sequence database and spectral library searches for improved peptide identification from tandem mass spectra.

Key Features:

• Parallel Processing with Hadoop: Uses the MapReduce paradigm to distribute peptide–spectrum matching tasks across nodes in a Hadoop cluster, enabling processing of tens of thousands of experimental spectra in hours.
• Hybrid Search Methodology: Integrates database search (consensus model spectra) and spectral library search (detailed intensity information) to enhance sensitivity and specificity in matching tandem mass spectra.
• Performance Improvements: Demonstrated substantial speedup and efficiency gains on a 400-core Hadoop cluster using environmental microbial community datasets relative to the serial MSPolygraph implementation.
• Error Rate Characterization: Employs a scoring metric rooted in statistical thermodynamics to provide conservative error-rate estimates when combining database and spectral-library results, with an estimated false discovery rate of 5%.

Scientific Applications:

• Global proteomics and large-scale peptide identification: Applied to large datasets to increase the number of spectra assignable to peptides by 57-147%, facilitating protein identification in complex samples such as environmental microbial communities.

Methodology:

Implements MapReduce parallelization on Hadoop; combines protein sequence database searches (consensus model spectra) and spectral library matching (intensity information) for tandem mass spectra; uses a scoring metric based on statistical thermodynamics for error estimation.