Automated Data-Driven Optimization of MRM Transitions and CollisionEnergies in LC–MS/MS

Importance of the topic

Targeted LC–MS/MS methods based on multiple reaction monitoring (MRM) are fundamental for sensitive, selective and quantitative analyses across clinical, pharmaceutical and environmental laboratories. Traditional MRM method development is labor- and sample-intensive because it requires prior knowledge of precursors, multiple injections, product ion scans and iterative collision energy (CE) tuning. Automated, data-driven MRM optimization addresses these bottlenecks by reducing operator dependence, minimizing sample consumption and accelerating development for high-throughput and complex workflows.

Objectives and overview of the study

• Develop and demonstrate an automated, data-driven workflow to generate optimized MRM transitions and collision energies with minimal injections and without prior precursor knowledge.
• Combine ultra-fast survey scans, retention-time scheduling and adaptive optimization algorithms to produce high-confidence MRM tables suitable for targeted quantitative analysis.
• Benchmark the approach using an Agilent sulfa drug standard mixture and compare injection counts and optimization efficiency against conventional MRM development.

Methodology and used instrumentation

• Sample: Agilent Sulfa drug standard mixture (P/N: 5190-0580) used as benchmark material.
• Instrument: Agilent 1290 Infinity LC coupled to an Agilent 6495D triple quadrupole LC/MS equipped with an intelligent on-board computer.
• Software and implementation: Algorithms and procedures implemented in Python 3.1 and within the Tune-Calibration-Diagnosis (TCD) environment.
• Key methodological elements:
  • Ultra-fast survey scans (rapid MS1/SIM-like acquisitions) to discover candidate precursor m/z values and retention times without prior lists.
  • Product ion surveys to find fragment m/z values for candidate precursors.
  • Reference normalization: at each acquisition time point the measured product-ion abundance is divided by the corresponding precursor-ion abundance at approximately the same retention time; this isolates fragmentation behavior from chromatographic and ionization variability and enables objective comparison of fragmentation responses.
  • Adaptive CE ramping and optimization windows controlled by algorithms to converge on per-transition optimum CE while minimizing unnecessary scans and injections.
  • Retention-time scheduling and multiplexing to parallelize acquisition and shorten total runtime.

Main results and discussion

• The automated workflow markedly reduced the number of required injections compared to the traditional approach. Where conventional development required multiple injections (several for RT finding, product ion scans and additional CE ramps), the data-driven workflow achieved rough precursor and fragment discovery in one sample run and CE optimization in a second required run (plus an optional blank for exclusion list generation).
• Reference normalization produced CE optimization curves by plotting relative product-ion abundance versus CE after normalization to precursor intensity; this enabled consistent peak reconstruction and objective CE selection even when absolute intensities varied because of chromatographic or ionization effects.
• Example optimized transitions (summary): several sulfonamide analytes produced stable optimized CE values in the ~-31 to -7 V range with varying relative fragment abundances (approx. 0.19–0.69 normalized heights). The workflow successfully identified primary product ions and associated optimum CEs for precursor m/z values including 247, 271, 279, 285 and 311 (individual numeric results reported in the poster table).
• Multiplexing of surveys and reference-normalized CE ramps increased throughput while maintaining robust selection of high-quality transitions. The approach removed the need for prior knowledge about analyte precursors, enabling discovery-driven MRM table generation for complex mixtures.

Benefits and practical applications

• Reduced method development time and operator load: fewer injections and automated parameter selection streamline routine and research method creation.
• Improved scalability: suitable for high-throughput laboratories and workflows analyzing large compound panels or unknown mixtures.
• Better reproducibility and objectivity: reference normalization decouples fragmentation optimization from chromatographic and ionization variability, reducing subjective manual tuning.
• Lower sample consumption: minimizes the number of runs required for complete MRM optimization, advantageous for scarce or costly samples.
• Ease of integration: implementation in Python and TCD suggests possible integration with existing vendor software and laboratory automation frameworks.

Future trends and possibilities for use

• Integration with machine learning: ML models could predict likely optimal CEs and fragment ions from survey data, further reducing experimental effort and improving accuracy across chemical classes.
• Extension to larger panels and non-targeted to targeted handoff: automated pipelines can automatically convert DIA or survey-driven discoveries into scheduled MRM assays for routine quantitation.
• Cloud-enabled and multi-instrument deployment: centralizing optimization results and models across instruments and sites can standardize MRM tables and accelerate method transfer.
• Real-time adaptive acquisition: closed-loop control where on-the-fly data guides immediate adjustment of acquisition parameters during the same run to optimize low-abundance analytes.
• Application to challenging sample matrices: improving deconvolution strategies and precursor isolation strategies will extend applicability to complex biological and environmental matrices with co-eluting isobars.

Conclusions

The presented automated, data-driven MRM optimization workflow efficiently identifies precursor/fragment pairs and optimal collision energies with minimal injections and no prior precursor lists. Reference normalization is a key innovation that separates fragmentation behavior from chromatographic and ionization effects, enabling objective CE optimization. Combined with multiplexing and retention-time scheduling, the approach shortens optimization timelines, lowers labor and sample needs, and supports scalable targeted LC–MS/MS method development suitable for high-throughput and complex analytical environments.

References

1. Agilent Technologies ASMS 2026 poster: Automated Data-Driven Optimization of MRM Transitions and Collision Energies in LC–MS/MS, Behrooz Zekavat et al., Agilent Technologies Inc., June 10, 2026.
2. US Patent Application No. 19/441,718, filed 01/06/2026 (related to automated MRM optimization concepts described in the poster).
3. Agilent promotional material and product pages (Agilent Technologies) cited in poster documentation.