Adapt modern causal learning models to high-dimensional perturbation data as occurring in single-cell genomics and combine this with methods from computational biology to model gene regulatory interactions.



The Helmholtz International Lab will focus on developing robust causal models in high dimensions,  learn deep representations of cell ensembles in response to perturbations, and add biological priors and interpret resulting trained models using experimental data such as epigenomics.


Governance, Management and Structure

The Helmholtz International Lab will be directed and represented by Fabian Theis, Yoshua Bengio, and Bernhard Schölkopf, who ensure alignment of work package goals with overall strategy and with the latest developments in the field.


Key Aims