Improved MS/MS Quality and Higher ID Rates Through Charge Tailored Collision Energies in dda-PASEF

Significance of the topic

Collision energy (CE) is a central acquisition parameter in data-dependent acquisition (DDA) proteomics workflows because it determines peptide fragmentation efficiency and therefore the quality of MS/MS spectra and downstream peptide identification. Standard CE settings are generally global and do not account for combined dependencies on precursor m/z, ion mobility and charge state. Tailoring CE to precursor-specific properties can produce more informative fragmentation, higher signal-to-noise, and increased identification rates—particularly important in challenging applications such as immunopeptidomics and deep proteome profiling.

Objectives and study overview

The study aimed to develop and evaluate charge-dependent collision energy optimization strategies for dda-PASEF on a timsTOF platform. Specific goals were to (1) generate an empirical CE database spanning m/z and ion mobility space, (2) derive continuous CE functions that incorporate precursor charge state and either m/z, ion mobility (1/K0), or both, (3) implement these functions in real time via the timsControl Python API for dynamic DDA acquisition, and (4) quantify impacts on MS/MS spectral quality and peptide identification rates versus the instrument default CE method.

Methods

• Data collection: 26 human digest files were acquired with varied CE settings covering m/z 100–1750 and ion mobility 0.65–1.35 Vs/cm² using dda-PASEF on a timsTOF (timsUltra AIP) platform.
• CE database generation: MS/MS outcomes across the acquisition space were compiled into a database to capture dependencies of optimal CE on m/z, ion mobility (1/K0) and precursor charge.
• Data analysis: Database searches were performed with SAGE; peptide- and spectrum-level metrics were extracted and further analyzed with custom Python scripts.
• CE optimization: Optimal CE values per bin were identified using boxplot statistics and 2D binning; model fitting produced continuous CE functions and enabled extrapolation across the precursor space.
• Implementation: Three charge-aware CE strategies (m/z-based, ion-mobility-based (1/K0), and combined m/z+ion mobility) were implemented via the timsControl Python API to assign precursor-specific CE in real time during dda-PASEF acquisition.
• Evaluation metrics: Fragmentation rate, SAGE hyperscore, matched fragment intensities, summed b- and y-ion intensities, matched intensity percentage, number of matched peaks, and MS2 intensity were compared between optimized and default CE acquisitions.

Instrumentation used

• timsTOF platform (timsUltra AIP configuration) for dda-PASEF acquisition.
• timsControl 7.0 and the timsControl Python API to enable real-time modification of DDA acquisition and precursor-specific CE assignment.
• SAGE search engine for peptide-spectrum matching and scoring.

Main results and discussion

• Three charge-dependent CE strategies were derived: (a) m/z-based, (b) ion mobility (1/K0)-based, and (c) combined m/z + ion mobility. Continuous CE surfaces were generated by 2D binning and model fitting and integrated into acquisition control.
• All three optimized strategies produced comparable improvements and outperformed the default ion-mobility-only CE method, yielding approximately 11% more peptide identifications across a human cell-line digest.
• MS/MS quality metrics improved systematically under optimized CE: fragmentation rate rose from 71% to 88%; SAGE hyperscore increased from ~63.1 to ~77.2; MS2 intensity increased from ~16,125 to ~54,794 (instrument-specific units); summed b- and y-ion intensities increased markedly (for example from ~2,799 to ~23,221 in representative comparisons); matched intensity percentage rose from 17% to 42%; and the number of matched peaks increased from 24 to 30 in the example summary table.
• Two-dimensional difference plots and manual spectral inspection showed consistent quality gains across m/z and ion-mobility space. Representative spectra with a delta CE of ~4.6 eV demonstrated higher signal-to-noise and clearer fragment ion series when optimized CEs were applied.
• The improvements were robust across charge states and precursor properties, indicating that incorporating charge state into CE selection is a beneficial general strategy for dda-PASEF.

Benefits and practical applications

• Higher peptide identification rates and improved spectral quality translate into greater sensitivity and confidence in proteomics experiments, which is especially valuable for low-abundance peptides and targeted fields like immunopeptidomics.
• Precursor-specific CE assignment enables more efficient use of instrument time by producing higher-information MS/MS spectra per scan, potentially reducing the need for repeated sampling of the same precursor.
• Real-time implementation via the timsControl Python API allows dynamic precursor scheduling and immediate incorporation of optimized CE functions into routine DDA workflows without offline post-processing.
• These strategies can be adapted for diverse sample types and experimental goals where charge-dependent fragmentation behavior is relevant (e.g., PTM analysis, peptide-centric assays, and complex mixture profiling).

Future trends and potential applications

• Adaptive and machine-learning-driven CE models: Training predictive models on larger, more diverse datasets could yield even finer-grained, precursor-tailored CE predictions that adapt to sample and instrument conditions in real time.
• Integration with search and quantification pipelines: Tighter coupling between acquisition optimization and identification engines could permit acquisition strategies that maximize identifications for underrepresented classes of peptides (e.g., modified peptides, short immunopeptides).
• Transferability to other instruments and fragmentation modes: Evaluating charge- and mobility-aware CE optimization across different platforms and dissociation methods will test generalizability and expand utility.
• Application to targeted and data-independent acquisition (DIA): Principles from charge-tailored CE optimization may inform collision energy scheduling for targeted MS/MS and DIA windows to improve fragment coverage.
• Real-time feedback loops: Implementing closed-loop systems where identification confidence influences subsequent CE adjustments could further increase throughput and identification depth.

Conclusion

Incorporating precursor charge state together with m/z and/or ion mobility into collision energy selection for dda-PASEF substantially improves MS/MS spectral quality and increases peptide identification rates. Three implemented strategies (m/z-based, 1/K0-based, and combined) produced comparable and consistent gains—approximately an 11% increase in peptide IDs and marked improvements across multiple spectrum-quality metrics. Real-time implementation via the timsControl Python API demonstrates practical viability for routine proteomics workflows and suggests immediate benefits for applications that depend on high-quality MS/MS data, such as immunopeptidomics.

References

• Lazear M. SAGE: An Open-Source Tool for Fast Proteomics Searching and Quantification at Scale. Journal of Proteome Research. 2023;22(11).