Speaker: Arjun Krishnan, PhD
University of Colorado Anschutz
About the Seminar
Talk Title: “Harnessing public omics data at scale: why it’s hard, what it reveals about disease biology, and what it teaches us about doing science”
Public repositories now hold millions of omics profiles, immune repertoire sequences, and biomedical texts—yet most of this data remains difficult to find, trust, or reuse. In this talk, I’ll describe how my lab tackles this challenge from three angles. First, we build machine learning and NLP tools (Txt2Onto and a growing ecosystem of metadata-annotation software) that convert messy, unstructured sample descriptions into standardized, searchable annotations. Second, we combine these annotations with molecular data to surface hidden biases in biomedical research itself, including a large-scale analysis quantifying sex bias across thousands of disease studies. Third, we show how archival data—11.5 million immune receptor sequences—can be repurposed with foundation models to predict multiple sclerosis-associated antibodies and disease risk. Throughout, I’ll emphasize the evaluation practices that make these predictions trustworthy, and close with what this work teaches us about doing rigorous, reusable, and equitable data-driven biomedical research.
Event Details
- Date: August 3, 2026
- Time: 11a.m. to 12 p.m.
- Location: Fifth and Halket Classroom