Mistle

Mistle constructs a searchable index of Prosit-predicted tandem mass spectra to enable efficient and accurate spectral-library searches in metaproteomics and proteogenomics.

Key Features:

• Spectral library predictions: Uses Prosit to generate peptide fragmentation predictions (predicted tandem mass spectra) for spectral-library construction.
• Indexing and efficient search algorithm: Builds a searchable index from Prosit predictions and identifies experimental mass spectra within the predicted library to support large search spaces.
• Deep learning integration: Leverages deep learning-based fragmentation models for authentic peptide fragmentation prediction.
• Performance comparison: Reports higher accuracy than database searches using MSFragger and improved runtime and memory usage versus other spectral library search engines, with a 4- to 22-fold decrease in RAM consumption.
• Applicability to large search spaces: Applicable to comprehensive sequence databases from diverse microbiomes and other extensive proteomic datasets typical in metaproteomics and proteogenomics.

Scientific Applications:

• Metaproteomics: Enables spectral-library searches across large, diverse microbial proteomes using Prosit-predicted spectra.
• Proteogenomics: Supports peptide identification in proteogenomic searches involving large or custom sequence databases.
• Environmental and community proteomics: Facilitates analysis of microbial communities and environmental samples with vast proteomic search spaces.

Methodology:

Generate spectral predictions using Prosit, construct a searchable index from those predictions, and match experimental tandem mass spectra against the index.