Research

Machine Learning for Single Cell Genomics

Methods for profiling the molecular content of individual cells at high throughput (collectively known as single cell genomics) provide a powerful and increasingly popular way for studying biology – from questions of basic science (e.g., how cells respond to certain stimulations) to translational applications (e.g., stratifying patient populations or screening for drug targets). Realizing this in practice, however, is challenging. The first challenge is conceptual – given the ever-growing richness of single cell assays (e.g., profiling different molecule types, employing labeling strategies), there is constant need for envisioning ways in which insight could be drawn. The second challenge is technical - the ensuing data is affected by a plethora of confounders, which make it difficult to process and to make sense out of. Our group is developing computational tools that build and extend upon advances in statistical machine learning and other disciplines to offer new ways to draw insight from single cell genomics and at the same time account for distortions in the data and evaluate our uncertainty in understanding it.

Leveraging sub-clonal dynamics for studying cancer and immunity

Evolution of cellular traits

Many processes in biology, including organ development, tumor growth, and adaptive immunity unfold over time in a way that originates from a small progenitor population and progresses through cell division and selective expansion of sub-clones. The emerging field of single-cell lineage tracing provides a unique opportunity to study these processes at an unparalleled resolution by tracking the clonal history of individual cells, while at the same time recording their transcriptomes. Our group is harnessing Cas9- based lineage tracing technologies to study clonal processes, with the current emphasis on tumor growth and its association with immunity. To enable these studies, we are developing the accompanying algorithms and software, targeting different aspects of the work such as inference of lineage trees and its theoretical guarantees, estimation of divergence times, analysis of heritable transcriptional programs, and analysis of transcriptional changes that take place during the evolution of clonal populations. This work brings together ideas from graph algorithms, statistical phylogenetics and generative modeling powered by deep learning.

Genomic dissection of immunity in its tissue context

Cells of the immune system routinely respond to cues from their environment and feed back to their surrounding in order to carry out their function. Our lab studies such processes by developing and applying computational tools that leverage assays for high throughput molecular and cellular profiling (including single cell genomics). Our current emphasis is on nascent techniques for high-resolution spatial analysis of RNA expression (collectively referred to as spatial transcriptomics). By combining these techniques with new computational tools (drawing on advances in statistical machine learning and computer vision) we seek to study how tumors achieve immune evasion by rendering tumor-infiltrating leukocytes inactive, how immune-epithelial cross talk changes during gut inflammation, and how the developing T cell repertoire is shaped locally in the thymus.