Improving real-time search through deep learning-enhanced fragment ion indexing
- 1University of Washington
Abstract
The fragment ion indexed database searching approach enables highly efficient tandem mass spectrometry database searches, including real-time search workflows. However, the index can require tens of gigabytes of memory, often exceeding the free memory available on instrument acquisition computers running the search. This restricts the size of the searchable database, the variable modification space, and the number of searches that can run concurrently. To address this problem, we developed an index-pruning pipeline within the Comet database search tool using Carafe, a deep learning tool that predicts fragment ion intensities based on peptide sequences. In our approach, only fragment ion positions that are predicted to be observable are written in the index, eliminating entries that are unlikely to contribute to spectrum detection. We evaluated this approach on two search spaces in three index configurations to demonstrate that masking significantly reduced the number of fragment ion index entries and, consequently, peak search memory use. Masked searches were also faster than unmasked searches and identified a comparable number of peptide-spectrum matches. Because all prediction and masking steps are performed before searching, their cost is amortized across repeated, future searches of the same database and modification configuration, introducing no per-spectrum search overhead. The complete implementation is open source with an Apache license.
Related
- STAMP as a topology-aware molecular language for fragment-level learning and editing
- Improving Molecular Docking Pose Recovery through Targeted Side-Chain Monte Carlo Sampling of an Apo Structure
- Improving beef quality through rumen microbial modulation by varying alfalfa hay proportions in the diet
- Improving allergic rhinitis symptoms through nutrition as an adjunct: a systematic review and meta-analysis on the efficacy of dietary supplements
- Improving the Usability of Donor Data from Registries Using an Extreme Learning Machine Approach to Upgrade Low-/Mid- to High-Resolution HLA Data
BibTeX
@misc{eng2026improving,
title = {{Improving real-time search through deep learning-enhanced fragment ion indexing}},
author = {J. Eng and M. Hoopmann and D. K. Schweppe and M. J. MacCoss and W. S. Noble and B. Wen},
howpublished = {bioRxiv},
year = {2026},
month = oct,
doi = {10.64898/2026.10.27.762387},
url = {https://www.biorxiv.org/content/10.64898/2026.10.27.762387v1},
note = {CC BY},
}
RIS
TY - UNPB
TI - Improving real-time search through deep learning-enhanced fragment ion indexing
AU - J. Eng
AU - M. Hoopmann
AU - D. K. Schweppe
AU - M. J. MacCoss
AU - W. S. Noble
AU - B. Wen
PY - 2026
DA - 2026/10/27
PB - bioRxiv
DO - 10.64898/2026.10.27.762387
UR - https://www.biorxiv.org/content/10.64898/2026.10.27.762387v1
AB - The fragment ion indexed database searching approach enables highly efficient tandem mass spectrometry database searches, including real-time search workflows. However, the index can require tens of gigabytes of memory, often exceeding the free memory available on instrument acquisition computers running the search. This restricts the size of the searchable database, the variable modification space, and the number of searches that can run concurrently. To address this problem, we developed an index-pruning pipeline within the Comet database search tool using Carafe, a deep learning tool that predicts fragment ion intensities based on peptide sequences. In our approach, only fragment ion positions that are predicted to be observable are written in the index, eliminating entries that are unlikely to contribute to spectrum detection. We evaluated this approach on two search spaces in three index configurations to demonstrate that masking significantly reduced the number of fragment ion index entries and, consequently, peak search memory use. Masked searches were also faster than unmasked searches and identified a comparable number of peptide-spectrum matches. Because all prediction and masking steps are performed before searching, their cost is amortized across repeated, future searches of the same database and modification configuration, introducing no per-spectrum search overhead. The complete implementation is open source with an Apache license.
ER -