November 4, 2015: Invited Speaker Seminar: Designing a System for Supervised Interactive Genomic Segmentation

Dr. Toby Dylan Hocking
 

Wednesday, November 4, 2015, 4:00 p.m.
Room EV003.309

Abstract

DNA copy number profiles characterize regions of chromosome gains, losses and breakpoints in tumor genomes. Although many models have been proposed to detect these alterations, it is not clear which model is appropriate before visual inspection the signal, noise and models for a particular profile. We discuss the design and implementation of the first computer vision system for genomic segmentation, SegAnnDB. It allows experts to visually inspect data sets and manually label some altered regions, which our system then uses to predict the precise alteration locations using supervised machine learning. I will discuss the quality of this system relative to previous unsupervised methods, and give a live demo of the system available online at http://bioviz.rocq.inria.fr/.

Biography

Dr Toby Dylan Hocking is a post-doctoral researcher in the McGill Human Genetics Department. His main research interests are on the interfaces of machine learning, data visualization, computer vision, and bioinformatics.