Bianca Dumitrascu: Understanding How Biological Matter Encodes Computation

Lorenza Favrot
January 11, 2023

The Department of Statistics and the Irving Institute for Cancer Dynamics welcome a new faculty member, Dr. Bianca Dumitrascu, Assistant Professor of Statistics and Herbert and Florence Irving Assistant Professor of Cancer Data Research (in the Herbert and Florence Institute for Cancer Dynamics).

A high school math puzzle competition and a visit to the campus of the Ecole Polytechnique in France set Dr. Bianca Dumitrascu on the path to computational biology. “Until then, I thought mathematics meant good puzzles and biology meant memorization,” Dr. Dumitrascu recalls. “During the visit, I met passionate students and faculty and learned about machine learning and bioinformatics, which were new to me. One of the students gifted me their bioinformatics textbook: it was written in French (a language I did not understand), but it sparked a deep curiosity.” When given the opportunity to pursue her undergraduate studies at MIT, she immediately coupled mathematics classes with undergraduate research in computational biology.

Earlier this month, Dr. Bianca Dumitrascu joined the Department of Statistics and the Herbert and Florence Irving Institute for Cancer Dynamics (IICD) as an Assistant Professor of Statistics and Herbert and Florence Irving Assistant Professor of Cancer Data Research. “Bianca’s research in the interface between machine learning and genetics as a way to understand the identity of single cells and their interactions in early development is directly relevant to the IICD’s mission. We look forward to welcoming Bianca in the New Year,” says Dr. Simon Tavaré, Professor of Statistics and Biological Sciences, and Director of the Irving Institute for Cancer Dynamics.

Prior to Columbia, she completed her PhD in computational biology at the Lewis Sigler Institute for Genomics (Princeton University) under the mentorship of Dr. Barbara E. Engelhardt. Her PhD research focused on the effect of experimental design in single-cell gene expression studies and on method development for structured, high-dimensional medical and genomic data. Then, she became a Fellow and Group Leader in the Computer Science Department at Cambridge University, a member of the School of Mathematics at the Institute for Advanced Study, and a visiting scholar in the Statistics Department at Duke University.

Her group will focus on understanding how biological matter encodes computation and whether we can mimic, predict, and control this computation. In development, cells integrate cues from their environments across spatial and temporal scales to build functional organisms, converting information into energy to make decisions in the presence of noise. “We will gain insight into these processes in the context of single-cell transcriptional and epigenetic profiling data, through techniques from interventional representation learning - such as multi-armed bandits, reinforcement learning, and dimensionality reduction - to statistical optimization and statistical physics,” describes Dr. Dumitrascu. She plans to work closely with experimental collaborators to implement robust, active, and transferable statistical inference pipelines which are end-to-end.

Dr. Bianca Dumitrascu thrives in interdisciplinary and collaborative environments. Columbia University, the IICD, and the Department of Statistics embody these qualities for her. “I move comfortably across different languages - physics, mathematics, computer science, genomics - and my interdisciplinary appointments offer me a unique opportunity to interact and collaborate with experts across all these fields,” explains Dr. Dumitrascu. “Furthermore, I always seek inspiration and motivation from my surroundings, and New York is a grounding, endless source of inspiration through its history, potential, and diversity. I look forward to interacting with the data science, biophysics, and computational biology communities at Columbia and beyond.”