Dr. Bob Pirok
Affiliation: Assistant Professor, Universiteit van Amsterdam, Netherlands
Talk Title: Automating Method Development and Data Analysis for Chromatography with Machine Learning
Bio: Bob Pirok obtained his PhD in 2019 in Amsterdam, defending two books of research, with the distinction cum laude and worked before at Shell. He is currently tenure-track assistant professor at the University of Amsterdam and focuses on the application of chemometrics to analytical chemistry with a special interest in method development and data analysis for multi-dimensional chromatography. Other areas of interest include retention modeling and (reaction) modulation techniques for LC×LC.
Pirok is visiting research professor at Gustavus Adolphus College in the group of Prof. Dwight R. Stoll. He is also visiting researcher in the group of Prof. André de Villiers of the Stellenbosch University in South Africa.
He received a number of international recognitions, including a Shimadzu Young-Scientist Award at HPLC2015 Beijing, the Young-Scientist-Award Lecture during the SCM-8 meeting in Amsterdam in 2017, the Csaba Horváth Young-Scientist Award at HPLC2017 Prague, the Journal of Chromatography Award during the ISCC Conference in Riva de Garda in 2018, and the SCM Award at the SCM-9 meeting in Amsterdam in 2019.
Pirok is the writer of the recently granted 1.2M€ NWO ENW-PPS TA PARADISE proposal coordinated by PI Prof. Arian van Asten. He was selected as Early Career Board member for the ACS journal Analytical Chemistry starting from 2021.
Abstract: Artificial intelligence is rapidly reshaping how we design, optimize, and evaluate analytical methods. In liquid chromatography-mass spectrometry (LC-MS), machine learning approaches, particularly Bayesian optimization (BO), offer the potential to revolutionize method development by accelerating decision-making, improving analytical Quality by Design (AQbD) practices, and enabling more systematic exploration of experimental space. Recent advances with the AutoLC platform have shown that fully closed-loop, unsupervised method development is not only feasible, but can operate robustly without human intervention, highlighting the growing maturity of AI-driven workflows.
At the heart of BO lies its kernel: the mathematical structure that encodes assumptions about chromatographic response surfaces and governs how the algorithm learns from data. Despite its importance, kernel choice is often treated generically, even though LC-MS method parameters follow well-defined physicochemical relationships that could be explicitly leveraged. In this talk, we will explore how different kernel families, default, ANOVA-style, stick-breaking, and entropy-search portfolios, shape optimization behavior in automated LC-MS workflows. We discuss how domain knowledge, chromatographic response functions, and mechanistic constraints may be incorporated into kernel design to better reflect realistic method-development landscapes. The presentation outlines a framework for evaluating kernel suitability and highlights how knowledge-guided kernels may enhance future AI-driven method development systems.
The automation of chromatographic method development is increasingly moving from expert-driven trial-and-error toward data-driven, self-optimizing workflows. While advances in instrumentation and optimization algorithms have made such automation technically feasible, its success ultimately depends on how effectively information can be extracted and interpreted from chromatographic data.
This challenge is particularly evident in comprehensive two-dimensional chromatography (LC×LC and GC×GC), where routine analysis can involve large series of samples and thousands of peaks per chromatogram. Manual comparison across runs is infeasible, making automated data interpretation a prerequisite rather than an optional add-on. In computational method development, where conditions are intentionally varied, the problem becomes even more complex and requires robust peak tracking across changing separation landscapes.
Data-driven method development strategies increasingly rely on retention models and machine-learning approaches to explore experimental parameter space efficiently, with goals ranging from improved selectivity to sustainability and cost efficiency. These approaches reduce, but do not eliminate, the need for explicit peak tracking by learning from structured chromatographic information. However, they expose a fundamental paradox: high-resolution separations are required to generate reliable training data, yet computational methods are themselves needed to guide the development of such separations.
In this presentation, we will outline our view on automated method development in chromatography and discuss recent advances from our group in retention-time alignment, peak tracking, and machine-learning-assisted data interpretation. Together, these developments illustrate how improved extraction and structuring of chromatographic information is enabling closed-loop, data-driven optimization and bringing comprehensive 2D chromatography closer to routine analytical practice.