Retention time dataset for heterogeneous molecules in reversed-phase liquid chromatography

Area of Science:

• Analytical Chemistry
• Computational Chemistry
• Cheminformatics

Background:

• Quantitative structure-property relationships (QSPR) are used to predict retention times in liquid chromatography (LC).
• Predicting retention times accurately is complex due to the influence of molecular structure and chromatographic method.
• Existing machine learning models for retention time prediction often lack transferability across different LC conditions.

Purpose of the Study:

• To develop a comprehensive dataset for improving the prediction of small molecule retention times in LC.
• To create more transferable models for retention time prediction across diverse chromatographic methods.

Main Methods:

• Compiled a dataset of over 10,000 experimental retention times for 343 small molecules.
• Included data from 30 different reversed-phase liquid chromatography (RPLC) methods.
• Utilized a wide range of chemical structures and common LC setups.

Main Results:

• The dataset encompasses a broad spectrum of small molecules and RPLC conditions.
• Provides a valuable resource for training and validating more robust retention time prediction models.
• Demonstrates the potential for enhanced model transferability with diverse experimental data.

Conclusions:

• A comprehensive dataset of experimental retention times has been established.
• This dataset facilitates the development of more accurate and transferable QSPR models for LC.
• Addresses the limitations of current models by incorporating method variability.