Evaluating Data Analysis Techniques for LC-IM-MS Data: Preprocessing, Untargeted Feature Finding, and DIA Fragmentation Alignment

Importance of the Topic

The combination of liquid chromatography (LC) with ion mobility–mass spectrometry (IM-MS) generates highly multidimensional data that enhance separation of complex samples and reduce chimeric spectra in data-independent acquisition (DIA). Efficient and accurate analysis of LC-IM-MS data is essential for lipidomics, proteomics, and environmental analyses such as PFAS profiling. Streamlined preprocessing and feature-finding algorithms enable reliable identification and quantification while supporting high throughput workflows.

Objectives and Study Overview

This study evaluates four-dimensional (RT, drift time, m/z, intensity) feature-finding algorithms and methods to align precursor and fragment ions in DIA data. Specific goals include:

• Comparing vendor (Agilent IMFE) and third-party (PNNL PreProcessor, MS-DIAL) feature finding implementations.
• Developing and testing two workflows for converting IM-MS data to DDA format for DIA fragmentation alignment.
• Assessing the impact of preprocessing steps (smoothing, saturation repair, polygon extraction) on feature detection and processing time.

Data from lipid extracts (NIST SRM 1950), BSA tryptic digest, and a PFAS sample were acquired on an Agilent 6560 Ion Mobility LC/Q-TOF with an Agilent 1290 LC system. Both single-pulse and multiplexed acquisition modes, as well as MS1 and mobility-aligned fragmentation (MAF) DIA approaches, were examined.

Used Instrumentation

The experimental setup included:

• Agilent 6560 Ion Mobility LC/Q-TOF mass spectrometer.
• Agilent 1290 series liquid chromatograph for separations.
• PNNL PreProcessor version 5.0 for preprocessing (CCS calibration, IM-to-DDA conversion, polygon extraction).
• Agilent IM-MS Browser (Ion Mobility Feature Extraction).
• Third-party software: MS-DIAL for lipid ID, MassHunter BioConfirm for peptide sequence coverage.

Methodology and Data Processing Strategies

Two main IM-to-DDA conversion workflows were compared:

• Workflow 1: Feature finding in IM-MS Browser followed by separate conversion to 3D DDA files.
• Workflow 2: Integrated feature finding and conversion within the PNNL PreProcessor, splitting chromatography regions into summed frames to preserve isomer resolution.

Key preprocessing enhancements introduced in PNNL PreProcessor 5.0:

• CCS calibration tab for automated tune-mix calibration via dataset matching or CSV input.
• IM-to-DDA tab combining feature extraction (IMFE) and 3D file generation in a single step.
• Polygon batch extraction to remove matrix background prior to feature finding.
• Three-point smoothing and saturation repair in both drift and chromatographic dimensions.

Main Results and Discussion

• Lipidomics: Workflow 2 leveraged high-resolution demultiplexing (HRdm) to increase triacylglycerol (TG) identifications in MS-DIAL, whereas Workflow 1 collapsed isomeric features, yielding no net gain. Splitting the chromatographic region into summed frames preserved HRdm benefits.

• Proteomics: A BSA digest processed via the IM-to-DDA workflow achieved 65% overall sequence coverage. Charge-state-specific coverage was 57% for +1, 30% for +2, and 10% for +3 precursors. Extraction window width comparison showed full width at half-max (FWHM) provided tighter windows (63% coverage) than full drift-time width (58%), suggesting reduced chimeric interference.

• PFAS Analysis: Polygon extraction prior to feature finding reduced processing time from 61 to 26 minutes across triplicate files. For targeted PFAS, preprocessing with smoothing and saturation repair recovered features that were missed or low-scored in raw data. Example perfluorotetradecanoic acid isomers were only detected after smoothing and single-ion feature allowance.

Benefits and Practical Applications of the Method

• Enhanced separation and confident feature detection in complex matrices.
• Automated preprocessing steps streamline high-throughput DIA workflows.
• Improved isomer resolution and reduced chimeric spectra through integrated IM-to-DDA conversion.
• Broad applicability across lipidomics, proteomics, and environmental contaminant analysis.

Future Trends and Applications

• Integration of machine learning for adaptive feature finding and interference correction.
• Real-time preprocessing and on-instrument calibration for faster decision making.
• Expansion to other omics fields (metabolomics, glycomics) and complex regulatory matrices.
• Development of standardized pipelines for cross-platform data interoperability.

Conclusion

The study demonstrates that advanced preprocessing features and an integrated IM-to-DDA workflow significantly improve feature finding, isomer resolution, and DIA fragmentation alignment for LC-IM-MS data. These enhancements support efficient, high-fidelity analysis across diverse chemical classes and will drive future innovations in multidimensional mass spectrometry.

Reference

1. Bilbao A. et al. Journal of Proteome Research. 2022;21(3):798–807.
2. Stow SM. et al. Journal of the American Society for Mass Spectrometry. 2024;35(8):1991–2001.
3. Tsugawa H. et al. Nature Methods. 2015;12(6):523–526.