Postdoc in Machine Learning for Oligopeptide Design (University of Groningen)

The advent of modern machine learning (ML) methodology is accelerating scientific progress by streamlining research across disciplines. In chemistry, it enables the modeling of complex structure-function relationships with unprecedented accuracy. Additionally, deep generative models allow for the generation of structures with desired function, thus enabling the design of novel molecules and materials.

Peptide-based materials provide attractive alternatives to materials derived from petrochemicals. By tuning the peptide sequence and modifications, material properties can be finely tuned. Their behavior at surfaces and interfaces are among the most fundamental, but also most important properties of small peptides. Control over these materials can enable a new generation of biodegradable materials, surfactants and coatings. However, how peptide sequence and modification affect the physicochemical properties and material performance remains unclear, let alone inverse design of these molecules.

The goal of this project is to develop ML models for: 1) predicting properties of oligopeptide materials based on peptide sequence and end-group functionalization; 2) guiding experimental structure optimization workflows to minimize the number of experiments to achieve desired properties; 3) generating new peptide sequences with desired target properties.
This position will be part of the 10 (https://bigchemistry.nl/), which combines laboratory automation with artificial intelligence. By building an autonomous ‘RobotLab’, large numbers of experiments can be carried out, yielding large datasets on properties of molecular systems. With this data, ML models are trained to predict the properties of molecular systems such as solubility, phase separation, critical micelle concentration, smell, toxicity, and reaction rates.