Unlocking Objective Numerical Evaluation of data analysis strategies: A Novel Platform to Generate Highly Realistic LC×LC and GC×GC data

Importance of the topic

Multi-dimensional chromatography techniques such as LC×LC and GC×GC generate complex two-dimensional datasets with overlapping peaks. Accurate and objective evaluation of data analysis workflows is critical for reliable quantitative results. However, experimental data lack a ground truth for comparison, while traditional simulations often fail to capture experimental nuances. A realistic simulation framework addresses this gap by combining experimentally derived parameters with known peak characteristics.

Study objectives and overview

This study presents a novel platform that generates highly realistic two-dimensional chromatograms for objective numerical evaluation of peak detection and quantitation methods. The key aims include:

• Defining a flexible peak model that incorporates height, width, skewness and modulation effects.
• Creating datasets with a fixed central reference peak and a moving test peak to assess recovery under interference.
• Comparing two common detection workflows across a broad parameter space.

Methodology

The simulation uses a skewed Lorentz-normal function parameterized by peak height, width (1D and 2D), shape, asymmetry and positional shift. Five parameters are systematically varied at multiple grid points, yielding over 70 unique chromatograms on a 100×100 data grid. Each chromatogram contains a constant central peak and a moving peak of predefined area to calculate recovery accuracy.

Two detection strategies are evaluated:

• Two-step approach combining one-dimensional peak detection with clustering to reconstruct full two-dimensional peaks.
• Watershed segmentation using image-based algorithms for direct two-dimensional peak delimitation.

Main results and discussion

Recovery performance maps illustrate the influence of peak asymmetry and modulation shift:

• High peak asymmetry leads to consistent over- or underestimation of true areas depending on the detection workflow.
• Modulation shifts create baseline variations that can degrade accuracy if not properly accounted for.

Neither method uniformly excels; effectiveness depends on data-specific characteristics, emphasizing the need to match the workflow to the dataset’s profile.

Benefits and practical applications

The simulation platform provides:

• An objective benchmark for chromatography data analysis algorithms against known ground truth.
• Insights into algorithmic limitations under realistic conditions.
• Guidance for selecting and tuning methods in LC×LC and GC×GC laboratories.

Future trends and opportunities

Ongoing and future developments include integrating simulated mass spectrometry signals to extend applicability to hyphenated techniques, incorporating real-world noise and retention time fluctuations, and testing machine learning–based peak detection. This modular framework enables the development of automated, optimized data processing pipelines tailored to specific analytical scenarios.

Conclusion

The proposed realistic simulation framework bridges the gap between purely theoretical and experimental datasets, offering a nuanced environment for quantitative evaluation of two-dimensional chromatography workflows. By systematically varying key parameters, it highlights workflow-specific strengths and weaknesses, empowering analysts to choose and customize algorithms based on data characteristics.

References

• Niezen L. et al. Analytica Chimica Acta 1201 (2022) 339605.
• Milani N. et al. Analytica Chimica Acta 1312 (2024) 342724.