AI/ML Now: Keeping Elution Order Intact During LC Method Transfer

Retention times are useful for identifying small molecules in LC–MS, but they are notoriously hard to predict across laboratories. Even small changes in column, mobile phase, pH or instrument plumbing can shift them. Most existing models must be trained on data from the same system they will later predict for.

In this second episode of AI/ML Now, Sebastien Böcker discusses the group’s recent Nature Methods paper: “Times Are Changing but Order Matters: Transferable Prediction of Small-Molecule Liquid Chromatography Retention Times.¹ This research introduces “2-step,” a machine learning approach that splits the problem in two. A graph neural network first predicts a retention order index from a compound’s structure and the column and mobile-phase conditions. A small set of anchor compounds then converts that index into retention times on the user’s own system. No model retraining is needed, and the method outperforms approaches that were trained on the target system, according to Böcker.

In this episode Sebastian Böcker answers the following questions:

• Which chromatographic variables cause the biggest changes in retention order, and what should chromatographers watch for when transferring methods?

• How far can the model be pushed beyond standard reversed-phase conditions, for example to hydrophilic interaction liquid chromatography (HILIC)?

• How should laboratories integrate predicted retention times into routine LC–MS annotation workflows, and what data could they contribute to RepoRT to improve the model?

Watch the second part of the interview here:

Watch the first episode here:

AI/ML Now: Rethinking Retention Time Prediction using Machine Learning
https://www.chromatographyonline.com/view/ai-ml-now-rethinking-retention-time-prediction-using-machine-learning

Reference

1. Kretschmer, F.; Harrieder, E.-M.; Witting, M.; Böcker, S. Times Are Changing but Order Matters: Transferable Prediction of Small-Molecule Liquid Chromatography Retention Times. Nat. Methods 2026. DOI: 10.1038/s41592-026-03243-2.