pLink

pLink performs proteome-scale identification of cross-linked peptides from mass spectrometry data using a two-stage open search strategy and fragment indexing to enable high-speed, reliable cross-linked peptide identification.

Key Features:

• Two-stage open search strategy: Implements a two-stage open search approach for cross-linked peptide identification as described for pLink-2.
• Fragment indexing: Augments the search strategy with fragment indexing to accelerate spectrum-to-candidate matching.
• High-speed performance: Reports approximately 40× speed improvement over pLink 1 and 3–10× speed improvement compared to Kojak.
• Proteome-scale capability: Targets proteome-scale identification of cross-linked peptides from complex datasets.
• Dataset validation: Validated on simulated datasets, synthetic datasets, 15N metabolically labeled datasets, and entrapment databases.
• Precision and sensitivity: Demonstrates improved precision and sensitivity in systematic evaluations of cross-link search engines.
• Increased identifications: Re-analysis of published proteome-scale cross-linking datasets yielded up to 27% more cross-linked residue pairs.
• Systematic evaluation framework: Accompanies four distinct analysis methods for assessing the credibility of ten state-of-the-art search engines.

Scientific Applications:

• Proteome-scale cross-link mapping: Identification of cross-linked peptides across whole-proteome mass spectrometry datasets.
• Structural proteomics: Generation of cross-linked residue pairs for structural and interaction studies.
• Benchmarking of search engines: Comparative evaluation and credibility assessment of cross-linking search engines using simulated, synthetic, 15N-labeled, and entrapment datasets.
• Re-analysis of published datasets: Re-processing published proteome-scale cross-linking data to increase cross-link identifications.

Methodology:

Uses a two-stage open search strategy augmented by fragment indexing and evaluates performance using simulated, synthetic, 15N metabolically labeled datasets and entrapment databases with four distinct analysis methods assessing ten search engines.