CHIMERYS: An AI-Driven Leap Forward in Peptide Identification

Importance of the Topic

In bottom-up proteomics, chimeric tandem mass spectra, which contain fragments from multiple peptide precursors, represent over 40 percent of data acquired in data-dependent acquisition methods. Traditional search algorithms assume a one spectrum one peptide model, leaving valuable signal unused and limiting identification depth. CHIMERYS addresses this challenge by applying artificial intelligence driven deconvolution to interpret complex MS2 spectra, enhancing identification rates and quantitative accuracy.

Study Objectives and Overview

• Introduce CHIMERYS, a novel AI based, cloud native search algorithm integrated into Proteome Discoverer 3.0
• Benchmark its performance against Sequest HT and INFERYS rescoring workflows on diverse datasets including HeLa cell lysates, Arabidopsis thaliana, yeast, and body fluids
• Validate false discovery rate calibration using entrapment searches and in silico chimeric spectra systems

Methodology and Instrumentation

CHIMERYS leverages the deep learning framework INFERYS 2.0 for predicting peptide fragment ion intensities and retention times. It processes each MS2 spectrum without pre filtering, simultaneously evaluating all candidate precursors within the isolation window and allocating measured intensities in a single deconvolution step. False discovery rate control is performed at the peptide spectrum match level using Percolator. Cloud based parallel microservices enable scalable analysis.

Instrumentation Used

• Thermo Scientific Orbitrap Exploris 480 mass spectrometer
• Thermo Scientific Orbitrap Eclipse Tribrid mass spectrometer
• Thermo Scientific Vanquish NEO liquid chromatography system

Key Results and Discussion

• Peptide identifications doubled compared to Sequest HT, achieving identification rates above 80 percent and an average of two PSMs per spectrum
• The number of peptides per protein increased by approximately 2.5 fold on average
• In silico chimeric spectra simulations demonstrated over 90 percent sensitivity in recovering true precursors
• Entrapment FDR analysis confirmed accurate false discovery control across multiple workflows
• Wider isolation windows and shorter chromatographic gradients yielded higher throughput without sacrificing depth, enabling up to 75 percent more quantified proteins in reduced measurement time

Benefits and Practical Applications

CHIMERYS enhances proteome coverage and throughput, making it suitable for high complexity samples such as fractionated cell lysates and low complexity matrices like body fluids. Its integration in standard software platforms allows laboratories to optimize acquisition parameters for increased efficiency and improved quantitative confidence in label free and multiplexed experiments.

Future Trends and Opportunities

• Application to large scale clinical and biological studies requiring deep proteomic profiling
• Integration with data independent acquisition and emerging acquisition strategies
• Refinement of AI models for improved de novo peptide identification and post translational modification discovery
• Expansion of cloud based platforms for collaborative and large cohort proteomics

Conclusion

CHIMERYS represents a significant advance in tandem mass spectrometry data analysis by employing AI based deconvolution of chimeric spectra. It substantially increases peptide and protein identifications, optimizes instrument utilization, and provides reliable quantitative performance, thereby accelerating proteomics research and its practical applications.

References

• Zolg DP; Gessulat S; Paschke C; Frejno M; et al. INFERYS rescoring Boosting peptide identifications and scoring confidence of database search results. Rapid Commun Mass Spectrom. 2021;e9128. doi 10.1002/rcm.9128
• Dorfer V; Maltsev S; Winkler S; Mechtler K. CharmeRT Boosting peptide identifications by chimeric spectra identification and retention time prediction. J Proteome Res. 2018;17(8):2581-2589. doi 10.1021/acs.jproteome.7b00836
• The M; MacCoss MJ; Noble WS; Käll L. Fast and Accurate Protein False Discovery Rates on Large Scale Proteomics Data Sets with Percolator 3.0. J Am Soc Mass Spectrom. 2016;27(11):1719-1727. doi 10.1007/s13361-016-1460-7
• Müller JB; Geyer PE; Colaço AR; Mann M; et al. The proteome landscape of the kingdoms of life. Nature. 2020;582:592-596. doi 10.1038/s41586-020-2402-x
• Bian Y; Zheng R; Bayer FP; Kuster B; et al. Robust, reproducible and quantitative analysis of thousands of proteomes by micro flow LC–MS/MS. Nat Commun. 2020;11:157. doi 10.1038/s41467-019-13973-x