Seminar Title: "AI-Assisted Biophysics: From Protein Structures to Cellular Phenotypes"
When
Sept. 15, 2026, 3:30 – 4:45 p.m.
Where
Haury, Room 216
Presenter Details
Seminar Information
Our ability to understand biological systems increasingly depends on extracting meaningful information from complex, high-dimensional data across multiple scales, ranging from protein structure to cellular organization. Yet conventional workflows are often constrained by extensive sample preparation, complex labeling requirements, and analytical tools built around restrictive assumptions. In this talk, I will discuss how integrating machine learning with label-free optical and imaging approaches can overcome these limitations and enable new ways to interrogate biological systems. I will first describe our recent work using supervised machine learning to predict protein secondary structure directly from circular dichroism spectra. By integrating spectral features with structural and molecular descriptors, these models provide accurate structural predictions without relying on predefined assumptions underlying conventional analysis methods. I will then move to the cellular scale, introducing label-free cellular morphometry pipelines that combine phase-contrast and brightfield microscopy with machine learning. These approaches enable quantitative cell type discrimination, longitudinal analysis of live cell responses to external stimuli, and detection of subtle phenotypic changes associated with cellular microenvironment and population density. Together, these studies demonstrate how AI can transform complex, label-free signals into quantitative, biologically interpretable measurements across molecular and cellular scales.
Seminar Host
Megha Padi, MCB
Contacts
Whitney DeGroot