24^rd Human Proteome Organization
World Congress (HUPO 2025)
November 09-13 | Toronto, Canada

 Chimerys is a universal search algorithm for bottom-up proteomics that deconvolutes chimeric fragmentation spectra for peptide identification and quantification. It supports data-dependent (DDA), data-independent (DIA), and parallel reaction monitoring (PRM) workflows using accurate fragment ion intensity predictions. Previously limited to a small set of post-translational modifications (PTMs) and a single instrument vendor, we developed a new deep learning model that expands PTM coverage and supports data from multiple vendors. Improved scoring logic enhances peptide identifications and data completeness while maintaining strict false discovery rate (FDR) control. 

 

 Mass spectra from timsTOF instruments differ significantly from Orbitrap-derived spectra due to variations in detector sensitivity, ion routing, fragmentation hardware, and the presence of ion mobility. In dda- and dia-PASEF, collision energy is set based on ion mobility, directly affecting fragmentation patterns and requiring tailored prediction models. Deep learning models trained on Orbitrap data do not capture these instrument-specific features, leading to reduced identification accuracy when used with Chimerys. To address this, we developed a deep learning model trained specifically on timsTOF data, extending Chimerys 5’s spectrum-centric deconvolution capabilities to this hardware platform. 

 

Botom-up proteomics data analysis presents growing computational challenges due to increasing data volumes and fragmented workflows requiring manual intervention and infrastructure management. The widespread use of fast-scanning mass spectrometers worsens these issues, as local storage and compute resources often fall short. To address this, we present an updated, cloud-based proteomics platform that streamlines the entire workflow—from raw data ingestion to analysis. The platform integrates scalable storage, high-performance processing, structured result storage, and intuitive data exploration, reducing overhead and enabling reproducible, large-scale studies.