biostudies-literatureUnknownTariq UNIGMS NIH HHSDatabase search algorithms reduce the number of potential candidate peptides against which scoring needs to be performed using a single (i.e. mass) property for filtering. While useful, filtering based on one property may lead to exclusion of non-abundant spectra and uncharacterized peptides - potentially exacerbating the <i>streetlight</i> effect. Here we present <i>ProteoRift</i>, a novel attention and multitask deep-network, which can <i>predict</i> multiple peptide properties (length, missed cleavages, and modification status) directly from spectra. We demonstrate that <i>ProteoRift</i> can predict these properties with up to 97% accuracy resulting in search-space reduction by more than 90%. As a result, our end-to-end pipeline is shown to exhibit 8x to 12x speedups with peptide deduction accuracy comparable to algorithmic techniques. We also formulate two uncertainty estimation metrics, which can distinguish between in-distribution and out-of-distribution data (ROC-AUC 0.99) and predict high-scoring mass spectra against correct peptide (ROC-AUC 0.94). These models and metrics are integrated in an end-to-end ML pipeline available at https://github.com/pcdslab/ProteoRift.bioRxiv : the preprint server for biology2024.08.21.609035https://www.ebi.ac.uk/biostudies/studies/S-EPMC11370541biostudies-literaturePredicting peptide properties from mass spectrometry data using deep attention-based multitask network and uncertainty quantification.PMC11370541R35 GM153434Saeed FTariq UfalsePredicting peptide properties from mass spectrometry data using deep attention-based multitask network and uncertainty quantification.Database search algorithms reduce the number of potential candidate peptides against which scoring needs to be performed using a single (i.e. mass) property for filtering. While useful, filtering based on one property may lead to exclusion of non-abundant spectra and uncharacterized peptides - potentially exacerbating the <i>streetlight</i> effect. Here we present <i>ProteoRift</i>, a novel attention and multitask deep-network, which can <i>predict</i> multiple peptide properties (length, missed cleavages, and modification status) directly from spectra. We demonstrate that <i>ProteoRift</i> can predict these properties with up to 97% accuracy resulting in search-space reduction by more than 90%. As a result, our end-to-end pipeline is shown to exhibit 8x to 12x speedups with peptide deduction accuracy comparable to algorithmic techniques. We also formulate two uncertainty estimation metrics, which can distinguish between in-distribution and out-of-distribution data (ROC-AUC 0.99) and predict high-scoring mass spectra against correct peptide (ROC-AUC 0.94). These models and metrics are integrated in an end-to-end ML pipeline available at https://github.com/pcdslab/ProteoRift.2024-01-01T00:00:00Z2024 Aug2026-07-02T03:20:19.768Z2025-04-06T10:14:19.231ZS-EPMC113705413922918510.1101/2024.08.21.609035