06/08/18  Portland, Ore.

Unsupervised analysis of large transcriptomic compendia with neural networks

Speaker: Casey Greene, PhD
Chief Fellow
Assistant Professor, Perelman School of Medicine
University of Pennsylvania
Director, Childhood Cancer Data Lab, Alex's Lemonade Stand Foundation

Time: Friday, June 8, 2018
Location: Center for Health and Healing, Third Floor, Conference Room 1A&B
3303 S.W. Bond Ave.
Portland, OR 97239

Abstract: Our understanding of the world, and specific maladies, improves when scientists metaphorically go fishing. However, the term "fishing expedition" is primarily used as a pejorative when applied to scientific projects. I'll provide my perspective on why science could become more accurate and efficient once we rescue this term. Scientific fishing expeditions are valuable because they can take research in new directions. Looking at other people's data can provide researchers with a new perspective on their area of study. But analyzing individual experiments from other labs is difficult. Instead of single datasets, our lab builds "nets" that we and other scientists can use to go fishing in large collections of public data comprised of many experiments. We've put these to the test on a few recent expeditions, and I'll share one or two stories from our own work. Finally, I'll discuss why I think large collections of public data are a uniquely valuable resource for the process of scientific discovery.