A universal comparison study of chromatographic response functions

Area of Science:

• Analytical Chemistry
• Chromatography
• Method Development

Background:

• Chromatographic response functions (CRFs) are crucial for optimizing separation quality.
• Existing CRFs vary in their effectiveness for guiding optimization algorithms.
• The performance of CRFs is often evaluated using limited datasets or specific conditions.

Purpose of the Study:

• To conduct a large-scale in silico comparison of various CRFs for chromatographic separations.
• To evaluate CRF performance using only first-order chromatographic data, without spectral information.
• To assess CRFs' ability to guide optimization when the number of sample compounds is unknown.

Main Methods:

• In silico comparison of a comprehensive set of existing and novel CRFs.
• Evaluation of CRFs' ability to guide search-based optimization using simulated chromatographic data.
• Analysis of CRF performance based on separation power, peak asymmetry, noise levels, and column efficiency.

Main Results:

• CRFs that increase monotonically with the number of observed peaks significantly outperform others.
• CRFs based on discrimination factor or peak-to-valley ratio handle peak asymmetry better than Snyder resolution (Rs).
• Optimal CRF performance is achieved when column efficiency is near or slightly below that required for baseline separation.

Conclusions:

• CRFs prioritizing the number of observed compounds are superior for comprehensive optimization.
• Incomplete chromatographic information necessitates spectral data for accurate optimization on insufficient efficiency columns.
• The choice of CRF significantly impacts the success of automated chromatographic method development.