Deep Learning in Drug Discovery

We are organizing a Leiden Data Science Meetup on Thursday the 14th of March, starting at 16:00 on Deep Learning in Drug Discovery.

Deep learning has been increasingly applied to a wide variety of (bio)chemical challenges. Several inspiring speakers share their expertise and insights on the application of deep learning for drug discovery and deep dive into the application for affinity prediction, virtual screening and de novo drug design.

We'll finish with some drinks, a few bites and some networking. So sign up for the event and save the date.

--- AGENDA ---

16:00 - 16:05 Welcome and agenda

16:05 - 16:10 Introductions

1. Artificial Intelligence in Drug Discovery - Capturing Chemistry in Language

Speaker: Prof. dr. Gerard J.P. van Westen, Computational Drug Discovery, Leiden Academic Center for Drug Research, Leiden University.

Abstract: The presentation provides an overview of the research conducted by the computational drug discovery group in Leiden, emphasizing the utilization of machine learning and the integration of chemical and biological data. The speaker showcases previously published examples and concludes by discussing exciting new possibilities on the horizon.

2. Large-scale docking as a basis for machine learning: a kinase case study

Speaker: Assoc. Prof. Dr. Anthe P.A. Janssen, Leiden Institute of Chemistry, Leiden University.

Abstract: Predicting the biochemical affinity of drugs for their protein targets is a notoriously difficult but extremely relevant problem. Here, we will look at an approach that uses large-scale docking to create an ML-ready database of 3D complexes with accompanying literature biochemical data. I’ll show that this idea provides promising first results with very basic models, and leaves much to be explored.

3. AlphaFold meets de novo drug design: leveraging structural protein information in multi-target molecular generative models

Speaker: Andrius Bernatavicius, PhD Candidate LACDR/LIACS, Leiden University

Abstract: Recent advances in deep learning are rapidly expanding the frontiers of virtual screening and de novo drug design.

We present PCMol - a GPT-like generative model that uses the internal protein representations of AlphaFold for conditioning the model to generate novel compounds for thousands of different targets. We illustrate the effectiveness of this approach by investigating the properties of generated molecules and comparing it to other existing methods.

17:10 - 17:15 - Closing remarks

17:15 - Drinks and social event