Integrated tools for scripted automation of tuning, review and analysis pipelines

Significance of the topic

Targeted liquid chromatography–mass spectrometry (LC–MS) assays provide high-quality, analyte-specific measurement but grow in complexity as panel sizes increase. Automated, scripted optimization and review pipelines reduce manual burden, improve reproducibility, and enable exploitation of advanced instrument capabilities (e.g., MS2/MS3 modalities and multiple activation types). The work summarized here demonstrates how an automation framework (Acquisition Composer) coupled with existing tools (Skyline CLI, Python scripts) can accelerate method development, transfer, and quantitative performance optimization for both proteomic peptides and small-molecule drug panels.

Objectives and study overview

• Develop and demonstrate Acquisition Composer workflows that automate iterative method development steps: LC peak finding, MS2 and MS3 parameter optimization, and post-acquisition transition selection.
• Apply the platform to: a proteomic kit (ProteomEdge DE175), a 393-peptide plasma kit, and a 32-compound drugs-of-abuse panel, including short LC methods (5 min) and assay transfer between LC systems.
• Quantify the analytical benefit of optimized activation types and energies (HCD vs resonance CID) and MS3 vs MS2 modes for limits of quantitation (LOQs) and signal response.

Methods and analytical workflow

Acquisition Composer implements iterative pre-acquire, acquire, post-acquire loops integrated into LC–MS workflows. The pipeline used:

• Automated LC peak detection (Find LC Peaks module) to assign features using precursor m/z and optional retention time or spectral library information.
• Systematic variation of activation type and normalized collision energy (NCE) across multiple injections to define optimal MS2 and MS3 settings per analyte.
• Data extraction from raw files with Skyline CLI, leveraging spectrum filtering to separate data acquired under different parameter sets.
• Python scripts for downstream analysis, transition re-selection and determination of LOQs using dilution curve experiments.

Key experimental elements: dilution curves for LOQ estimation; repeated LC injections varying activation type (collision-cell HCD vs resonance CID) and energy; comparison of MS2 and MS3 acquisition modes; post-acquisition selection of optimal transitions to minimize LOQ.

Used instrumentation

• Stellar MS mass spectrometer (Thermo Scientific) with MS2 and MS3 capability, supporting collision-cell HCD and resonance CID activation types.
• Vanquish Neo and Vanquish Horizon LC systems used for method transfer and routine acquisition.
• Evosep LC system (original method reference) for comparison of retention time trends.
• Columns referenced: ES75 150 µm x 150 mm and comparable ES75150PN.
• ProteomEdge Discovery Edge DE175 kit (96-well plate of qRePS heavy-labeled peptides) and other reference peptide panels.
• Software: Acquisition Composer (automation framework), Skyline CLI for data extraction, and Python for data analysis and decision logic.

Main results and discussion

• Peptide assays (DE175 and related kits):
  • Optimal MS2 settings: resonance CID NCE ~30% with 2 ms activation time and q = 0.25; collision-cell HCD optimal near ~25% NCE.
  • HCD provided higher signal for ~62% of peptides; however, overall peptide activation parameters required little per-peptide optimization.
  • Choosing the best MS2 or MS3 condition per peptide improved LOQ by ~2.5-fold relative to default MS2 HCD.
• Small-molecule drugs (32-compound panel):
  • Optimal collision energies for small molecules varied substantially between compounds; the default HCD of 30% was suboptimal in many cases.
  • Reducing HCD to ~22% produced on average ~35% higher signal compared to the 30% default.
  • MS3 offered substantial benefit: selecting the best MS3 modality per analyte produced approximately 6.3-fold lower LOQs compared to MS2 HCD for the drugs-of-abuse panel.
  • Overall, for the drug panel MS3 gave LOQs that were the same or better than MS2 for 84% of molecules, and strictly better for 48% of molecules; for peptides the comparable figures were 70% (same or better) and 40% (better).
  • Activation-type performance split varied by analyte: MS2 optimal activation was roughly balanced between HCD and CID, while MS3 often showed best signals with HCD/HCD or CID/CID sequences; when MS3 strongly outperformed MS2, the first activation was frequently CID.
• Assay transfer: Moving the ProteomEdge assay from Evosep to Vanquish systems was straightforward using expected retention-time trends; a few outliers were attributed to missed-cleavage peptides not present in the Thermo datasets.
• Post-acquisition transition optimization (i.e., choosing the best fragment ions after data were collected) further reduced LOQs, illustrating the value of decoupling acquisition from final transition choice when instrument and software allow it.

Benefits and practical applications of the method

• Scalability: Automation enables routine development and optimization for large targeted panels (hundreds of peptides or dozens of small molecules) with reproducible decision logic.
• Improved quantitative performance: Per-analyte optimization of activation type and energy, plus MS3 options, materially lowers LOQs and improves selectivity for challenging analytes (notably certain small molecules such as zolpidem in the study).
• Faster assay transfer: Automated retention-time alignment and LC peak finding facilitate method migration between LC platforms with minimal manual tuning.
• Workflow integration: Combining instrument control, Skyline-based extraction, and scripted analysis enables closed-loop iterative refinement (pre-acquire/acquire/post-acquire) and supports expert-review or QC-triggered actions in routine use.

Future trends and potential applications

• Software/UI development: Building a user-facing interface around Acquisition Composer will make iterative optimization workflows accessible to laboratory users without scripting expertise.
• Per-analyte MS3 tailoring: Extending MS3 optimization to per-peptide or per-compound settings and exploring wider Q1 isolation windows for MS3 could further enhance sensitivity and selectivity.
• Automation at scale: Applying these scripted loops to very large panels (thousands of targets) and integrating machine-learning based parameter prediction from prior datasets can further reduce experimental iterations.
• Integrated QC and decision logic: Implementing automated QC triggers, expert-review workflows and automated transition choice pipelines will streamline regulated laboratory deployments.
• Broader adoption: The pipeline approach can be generalized to other instruments and vendors where programmatic control and robust data extraction are available, promoting harmonized multi-site assays.

Conclusions

The Acquisition Composer platform demonstrates that scripted, iterative optimization of LC–MS parameters can substantially improve quantitative performance and throughput for targeted assays. For peptides, modest per-analyte gains are achievable (≈2.5× LOQ improvement when selecting optimal MS2/MS3 settings), while small molecules can benefit more dramatically from per-analyte activation optimization and MS3 strategies (≈6.3× LOQ improvement observed for a drugs-of-abuse panel). Coupling instrument automation, Skyline extraction, and scripted analysis enables efficient assay transfer, objective parameter selection, and post-acquisition transition optimization, supporting high-throughput targeted proteomics and small-molecule panels in research and applied laboratories.

References

1. Remes P.M., et al. Hybrid quadrupole mass filter–radial ejection linear ion trap and intelligent data acquisition enable highly multiplex targeted proteomics. Journal of Proteome Research. 2024;23(12):5476–5486.
2. Kotol D., et al. Targeted proteomics analysis of plasma proteins using recombinant protein standards for addition-only workflows. Biotechniques. 2021;71(3):476–483.