About the team you will join

As a Machine Learning Research Intern, you’ll join a team of engineers and researchers building algorithms to improve and accelerate our internal drug discovery pipeline. You will be working in the series-expansion team, composed of 3 ML engineers. On a day-to-day basis, you will interact with Victor Saillant.

Your role

You will explore the topic of molecular generation in depth and be responsible for literature review, implementation and training/evaluation of models on public and proprietary data. 

Your internship should last between 4 and 6 months, and can start as early as possible in 2024.

Subject of the internship

The objective of the internship is to address the problem of molecule generation conditioned on a protein, and possibly in a constrained chemical space and additional physico-chemical properties. The proposed method involves the use of diffusion models on graphs to address this issue (see references [1][2]). Additionally, alternative approaches, like auto-regressive models, may be explored in a subsequent phase (see references [3][4]).

• You are a Masters student or a PhD student in Computer Science, Applied Mathematics, Bioinformatics, or a related field.

• You are actively interested in the field of machine learning, and enjoy keeping up to date with current developments.

• Your knowledge of mathematics and statistics allows you to understand and critically evaluate research papers from the field.

• You are comfortable with Python as a programming language, and ideally have hands-on experience with the implementation (using PyTorch/Jax/Tensorflow), training, and evaluation of deep learning systems.

• You are curious and eager to spend time learning new topics from people with diverse backgrounds, and believe that machine learning can play a pivotal role in biology and chemistry for drug discovery.

Nice to have

• Experience in representation learning, generative modeling.

• You’re interested in complex structured data such as graphs, point clouds, and text.

• Knowledge in biology and/or chemistry/chemoinformatics is a strong plus.

References

• [1] Huang, L., Xu, T., Yu, Y. et al. “A dual diffusion model enables 3D molecule generation and lead optimization based on target pockets”. Nat Commun 15, 2657 (2024). https://doi.org/10.1038/s41467-024-46569-1

• [2] Schneuing, Arne, Yuanqi Du, Charles Harris, Arian Jamasb, Ilia Igashov, Weitao Du, Tom Blundell et al. “Structure-based drug design with equivariant diffusion models.” arXiv preprint arXiv:2210.13695 (2022).

• [3] Zhung, W., Kim, H. & Kim, W.Y. 3D molecular generative framework for interaction-guided drug design. Nat Commun 15, 2688 (2024). https://doi.org/10.1038/s41467-024-47011-2

• [4] Alexander S. Powers, Helen H. Yu, Patricia Suriana, Rohan V. Koodli, Tianyu Lu, Joseph M. Paggi, and Ron O. Dror. Geometric Deep Learning for Structure-Based Ligand Design ACS Central Science 2023 9 (12), 2257-2267 DOI: 10.1021/acscentsci.3c00572 https://pubs.acs.org/doi/full/10.1021/acscentsci.3c00572