The GREYC (computer science) and CERMN (pharmacology) research laboratories have been working for many years in a joint group to develop innovative computer methods for processing data relating to drugs and, in particular, to try to predict and understand how a new molecule acts from knowledge extracted from a set of molecules.

Since molecules can be modelled by graphs, it is possible to train a GNN (Graph Neural Network) to classify or predict the action of molecules, but currently no explanation of the network's decision-making is given. The explicability of networks is fundamental in predicting the therapeutic properties of molecules. Indeed, it allows

• to validate predictions before committing efforts and resources to the synthesis of a molecular series
• to gain insight into the key physico-chemical properties that a molecule must possess in order to have a targeted biological action.

The aim of this thesis is to propose solutions to explain the decisions made by a neural network operating on graphs with a view to applying it to data on chemical molecules. No knowledge of therapeutic chemistry is required.

Funding: €2,100 gross, see https://www.enseignementsup-recherche.gouv.fr/fr/le-financement-doctoral-46472

Location: University of Caen Normandy, GREYC laboratory, CODAG team
Contact: jean-luc.lamotte@at@unicaen.fr
Application deadline: 13 May 2024
Detailed subject:

• In English,
• En Francais

Application: CV, transcript of Bachelor, MSc (or equivalent level) marks, covering letter explaining how your application matches the subject, and contact details for one or more referees.

Candidate profile: the candidate must be enrolled in the final year of a Master's degree or an engineering degree, or hold such a degree, in a field related to computer science or applied mathematics, and have solid programming skills. Experience in data science, deep learning, etc. would be a plus. The candidate must be able to write scientific reports and communicate research results at conferences in English.