IntroSpect

IntroSpect improves HLA-binding peptide identification from immunopeptidomic mass spectrometry data by learning peptide motifs from initial high-confidence hits to construct motif-guided targeted databases that enhance sensitivity and reduce false positives.

Key Features:

• Motif-Guided Database Building: Learns peptide motifs from high-confidence hits in an initial database search and constructs a targeted database based on those motifs.
• Improved Sensitivity and Accuracy: Demonstrated across three HLA class I datasets with an average sensitivity improvement of 80% over conventional shotgun proteome searches with non-specific enzymatic digestions and ~96% accuracy validated by synthetic experiments.
• Independence from External HLA Data: Operates without external HLA data, unlike SpectMHC, enabling application to both well-studied and poorly characterized HLA types.
• Controlled False Discovery Rate (FDR): Maintains a global FDR that can be controlled similarly to traditional database search engines.
• Practical Application in Neoantigen Discovery: Has been applied to discover neoantigens directly from mass spectrometry data to support cancer immunotherapy and personalized medicine.

Scientific Applications:

• Immunopeptidomics: Improves peptide identification in immunopeptidomic mass spectrometry studies, refining analysis of presented peptides.
• Cancer Research: Facilitates discovery of neoantigens from mass spectrometry data to support development of personalized cancer vaccines and therapies.
• HLA Typing Studies: Enables analysis across diverse HLA types without requiring external HLA information.

Methodology:

Performs an initial conventional database search, identifies high-confidence peptide hits, extracts peptide motif information from those hits, constructs a motif-guided targeted database, and conducts a secondary search against the tailored database to enhance sensitivity and reduce false positives.