Workshop: AI for DNA, Chemicals, and Microbiome - Models, Mechanisms, and Applications
Microbiome data serves as a central application area, highlighting challenges in sparse, compositional, and heterogeneous data, and motivating generalizable modeling approaches. The aim with this workshop is to map the current state of the field, identify key challenges in modeling and evaluation, and foster new collaborations.
This workshop brings together researchers in machine learning, biology, and chemistry to advance AI methods for molecular and biological systems. It focuses on foundation models trained on genomic and molecular data, as well as multimodal approaches for integrating omics data. We will also dive into a novel field of mechanistic interpretability to understand how well models capture biological processes.
Program
| 09:30 | Morning coffee |
| 10:00 |
Talk on DreaMS: a Foundation Model for Tandem Mass Spectrometry by Roman Bushuiev, PhD student at Czech Institute of Informatics, Robotics and Cybernetics |
| 10:30 |
Talk on DNA Foundation Models by Frederikke Isa Marin, Postdoctoral Researcher at University of Copenhagen |
| 11:00 |
Talk on Proteome-Augmented Metabolomics Improves Disease Risk Prediction in Population Cohorts by Dewei Hu, PhD Student at University of Copenhagen |
| 11:30 | Lunch |
| 12:30 |
Talk on Opening the Black Boxes of Biological AI: Mechanistic Interpretability of Single-Cell Foundation Models by Ihor Kendiukhov, Founder at Biodyn |
| 13:15 |
Talk on GEA: Graph Explainable Attribution for decomposing GNNs using Sparse Autoencoders by Edir Sebastian Vidal Castro, Data Scientist at Biotechnology Research Institute for the Green Transition (BRIGHT) |
| 13:45 | Afternoon coffee |
|
14.15 |
Talk on MassSpecGym: Benchmarking the Discovery and Identification of Molecules from Mass Spectra by Anton Bushuiev, PhD Student at Czech Institute of Informatics, Robotics and Cybernetics |
|
14.45 |
Talk on Data-driven Approaches to Understanding Asthma in Childhood by Shiraz Shah, Senior Researcher at Copenhagen Prospective Studies of Asthma in Childhood, Gentofte Hospital |
| 15:15 |
Talk on Foundation Models in Practice: Proteomics and Metabolomics by Damian Rafal Plichta, Head of Data Science at Novonesis |
| 15.45 | Concluding remarks and thanks for today! |