PepArML

PepArML integrates multiple peptide-search engines to increase the sensitivity and accuracy of peptide identification from tandem mass spectra for proteomics analyses.

Key Features:

• Integrated search-engine support: Integrates Mascot, Tandem (native, k-score, s-score), OMSSA, MyriMatch, and InsPecT (utilizing MS-GF spectral probability scores) for combined analysis.
• Dynamic search configuration: Dynamically reformats spectral data and constructs engine-specific search configurations on-the-fly.
• Machine-learning result combiner: Applies an unsupervised, model-free machine-learning combiner to select optimal peptide identifications per spectrum using enzymatic digestion patterns, retention time, precursor isotope clusters, mass accuracy, and proteotypic peptide properties.
• False-discovery rate estimation: Estimates false-discovery rates to provide confidence metrics for identifications.
• Scalable computation: Leverages cluster, grid, and cloud computing resources for large-scale peptide identification tasks.
• Standardized output: Emits peptide identifications in pepXML format.

Scientific Applications:

• Increased identification sensitivity: Identifies two- to three-fold more spectra than individual search engines at a 10% false-discovery rate.
• Proteomics mapping: Supports protein function, interaction, and modification mapping in large-scale mass spectrometry proteomics analyses.

Methodology:

Runs Mascot, Tandem (native, k-score, s-score), OMSSA, MyriMatch, and InsPecT (MS-GF spectral probability scores); dynamically reformats spectra and constructs engine-specific search configurations; combines engine outputs with an unsupervised, model-free machine-learning combiner evaluating enzymatic digestion patterns, retention time, precursor isotope clusters, mass accuracy, and proteotypic peptide properties; estimates FDR; outputs pepXML; and utilizes cluster, grid, and cloud computing for processing.