Our research goal is immuno-engineering: to alter the behavior of specific immune cell populations in disease contexts (autoimmune disease, organ transplant, and cancer) without compromising the body’s homeostatic immune function (e.g. defense against pathogens). Thus, it is critical to develop a nuanced understanding of how different immune cells sense and respond to environmental cues across the body, in both physiological and disease settings. To this end, our lab’s major focus is reverse-engineering the underlying logic of immune cells (molecular networks that drive cellular responses) from high-dimensional molecular measurements of immune cells in action (sensing and responding to perturbations, disease conditions, etc.).

The lab’s focus is transcriptional regulatory network inference, modeling gene expression as a function of transcription factor activities, from gene expression and measurements of chromatin state. Chromatin accessibility measurements by ATAC-seq, together with transcription-factor DNA-binding preferences (motifs), can be used to broadly profile potential transcription factor binding events in relatively small populations of cells. Thus, we have used ATAC-seq with RNA-seq to enable de novo inference of transcriptional regulatory networks in physiological settings where sample material is limiting (e.g., intestinal immune cells in response to microbial/genetic perturbations). To date, most of these efforts have been in mouse models, as it is generally not possible to obtain sufficient sample material for similar experimental designs in human. To build human immune cell models, we are developing multi-task learning approaches to leverage evolutionarily conserved relationships and borrow statistical power from mouse datasets for inference in human. We are also developing methods to enable transcriptional regulatory network inference in very rare cell populations, as measured from single-cell RNA-seq experiments.