Schmidt Center - MIT EECS Colloquium, featuring Ben Raphael (Princeton University)
About this Event
415 Main Street, Cambridge, MA 02142
https://www.ericandwendyschmidtcenter.org/events/katherine-hellerPlease note that due to travel issues, this event has been postponed. We hope to reschedule but do not have details at this moment. We look forward to seeing you at a future Schmidt Center event!
Schmidt Center - MIT EECS Colloquium: Integrative inference of tissue architecture across space, time, and modality by Ben Raphael
Monday, November 3, 2025
4:00 - 5:00 pm (refreshments at 3:30 pm)
Broad Auditorium (Merkin building, 415 Main St.)
✍️ Learn more
Please join us for a colloquium featuring Ben Raphael, Graduate Class of 1991 Professor of Computer Science, Princeton University, on Integrative inference of tissue architecture across space, time, and modality.
This colloquium is part of a series hosted jointly by the Eric and Wendy Schmidt Center at the Broad Institute and the Department of Electrical Engineering and Computer Science at MIT. Ben's colloquium will run from 4:00-5:00 pm with refreshments served at 3:30 pm.
The colloquium will be held at the Broad Institute Auditorium (415 Main St) as well as virtually via YouTube Livestream: broad.io/ewsc.
If you do not have a Broad badge, please show up at the 415 Main Street entrance 10 minutes prior to the event to be escorted to the talk.
Questions? Email Amanda Ogden at aogden@broadinstitute.org.
Abstract:
The spatial organization of tissues is essential to their biological function. Recent spatialomics technologies measure mRNA, protein, metabolite, and other modalities at thousands of locations within tissue sections, revealing spatial patterns of cell types and molecular activity. However, current datasets are often sparse and incomplete due to technological and cost constraints. I will present machine learning approaches to overcome this sparsity by modeling the latent geometry of individual tissue slices and by integrating measurements across multiple modalities over space and time. These approaches utilize deep neural networks and novel formulations of low-rank optimal transport. We apply the resulting methods to analyze spatial variation in cell types and gene expression in normal tissues, derive gene expression gradients within tumor microenvironments, reconstruct three-dimensional tissue architecture across modalities,and describe spatiotemporal changes in expression during organismal development.