Computational Morphogenomics Group Dumitrascu Lab @ Columbia University


You can learn more about ongoing specific projects below.

Explainability + Genomics

Dimensionality Reduction Algorithms with a focus on explainability for single cell transcriptomics experiments.


Physics + ML

Physics informed machine learning tools for understanding self-organizing systems. Multi-agent simulations, graph neural networks and symbolic regression models for in silico active matter, developmental patterns and ecology.

In vitro vs in vivo systems

Transfer learning and multi-modal learning for organoid to scRNA-seq integration and comparison. What gene regulatory signatures encountered in vivo are recapitulated in organoid cultures?

Mechanics + Transcriptomics

Bayesian machine learning methods for inferring mechanical properties and morphological features of cellular aggregates with the goal of quantifying their interaction with transcriptomics.

Infer + Perturb

Experimental design and multi arm bandits for optimal perturbations in transcriptomic studies.