Jishnu Das, PhD

Assistant Professor of Bioinformatics, Department of Immunology and Director, Computational Immunogenetics Core, University of Pittsburgh

“Studying Immunomodulation in Infectious and Autoimmune Disease Using Interpretable Machine Learning”

Rapid technological advances have generated numerous deep cellular and molecular omics profiles in health and disease. However, gleaning meaningful insights from these datasets requires complex analytical approaches. We research the development and use of novel statistical, machine learning and network systems approaches to analyze high-dimensional datasets and elucidate mechanisms of immune regulation and dysregulation. I will present my lab’s efforts to integrate multi-scale, multi-modal datasets using Significant Latent Factor Interaction Discovery and Exploration (SLIDE)—a first-in-class interpretable machine learning technique for identifying significant interacting latent factors underlying outcomes of interest from omics datasets. While other machine learning methods provide correlative insights alone, SLIDE provides inference of actual biological mechanisms that other methods do not offer. SLIDE has been applied to a wide range of infectious, autoimmune and alloimmune diseases to gain mechanistic insights into complex regulatory architectures. I will also describe Sliding Window Interaction Grammar (SWING), a novel interaction language model that uses deep learning to decipher the lexicon of biological interactions. SWING is a generalizable zero-shot approach that learns the language of peptide/protein interactions, their immunomodulation by pathogens, and genetic variation in health and disease.