Deep Learning in Life Sciences - Edition 2026

As part of the Blended Intensive Program (BIP) action within the 4EU+ alliance of European universities, Heidelberg University and Sorbonne Université will coordinate a joint course on Applications of Deep Learning in Life Sciences (DeepLife) involving the universities of Paris-Sorbonne, Warsaw, Prague, Milano and Heidelberg. The purpose of this course is to give an insight in the very active field of DL in the field of biomedicine and highlight recent examples of applications.

It is a follow-up of the Seed4EU+ initiative of the same name that can be found here: https://deeplife4eu.github.io/ .

Organization

This course will be organized in 7 weeks of advanced Journal clubs (start: week of 9.02.2026)

First date: Thursday 12 at 4:30 pm to 5:30 pm.

The journal clubs will be based on the online lectures and material that can be found here: Lectures.

The journal club sessions will be 60-minute and online. They will be animated by teachers from all participating universities. Four groups of 2 students from the same university will participate in each session:

1.Background team: the group will explain the context and background of the paper (15 minutes).

2.Presenter team: the group that will present the paper (20 minutes).

3.Reviewer team: the group that will act as reviewers and ask questions. Please send your question 2 days before the journal club so they can be shared to other groups (25 minutes).

4.Writer team: the group that will write a 2-page summary of the paper and the session, with up to 10 references.

Here is the list of papers, for which you will have to constitute groups:

1. Protein embeddings and deep learning predict binding residues for various ligand classes. Littmann, M., Heinzinger, M., Dallago, C., Weissenow, K., & Rost, B., Scientific Reports 2021.
   https://www.nature.com/articles/s41598-021-03431-4
2. Comparative evaluation of methods for the prediction of protein–ligand binding sites. Utgés, J. S., and Barton, G. J., Journal of Cheminformatics 2024.
   https://link.springer.com/article/10.1186/s13321-024-00923-z
3. Ab initio characterization of protein molecular dynamics with AI2BMD. Wang, T., He, X., Li, M. et al., Nature 2024.
   https://www.nature.com/articles/s41586-024-08127-z
4. VAMPnets for deep learning of molecular kinetics. Andreas Mardt, Luca Pasquali, Hao Wu and Frank Noé, Nat Commun 2018.
   https://www.nature.com/articles/s41467-017-02388-1
5. Deep Generative Markov State Models. Hao Wu,Andreas Mardt,Luca Pasquali and Frank Noé, Advances in Neural Information Processing Systems 2018.
   https://proceedings.neurips.cc/paper_files/paper/2018/file/deb54ffb41e085fd7f69a75b6359c989-Paper.pdf
6. De novo design of protein structure and function with RFdiffusion. Watson, J.L., Juergens, D., Bennett, N.R. et al., Nature 2023
   https://www.nature.com/articles/s41586-023-06415-8

The associated lectures (see here Lectures) associated to each papers are:

• 1 → Deep learning models for protein-ligand binding site prediction, David Hoksza (Prague)
• 2 → Deep learning models for protein-ligand binding site prediction, David Hoksza (Prague)
• 4 → Deep Architectures for sampling macromolecules, Grégoire Sergeant-Perthuis
• 5 → Deep Architectures for sampling macromolecules, Grégoire Sergeant-Perthuis
• 6 → Protein design in the deep learning era, from inverse folding to diffusion models, Elodie Laine

Contact person for information

Prof. Dr. Carl Herrmann – IPMB carl.herrmann@uni-heidelberg.de
Prof. Grégoire Sergeant-Perthuis - CQSB gregoire.sergeant-perthuis@sorbonne-universite.fr