Barber, Bradley,Melina Diamond-Sagias, Yunsong Wang, Jonathan Hillman, Zachary Wilson-Long

Title:
PeakLearner, an interactive web app for machine learning in genomic data

Abstract:
The advent of DNA sequencing was a significant milestone in medical research and has led scientists to better understand the molecular foundations of many diseases as well as identify over 6,000 genetic disorders. Recently there have been drastic improvements in the efficacy, cost, and speed of sequencing such that it is now the computing capabilities that are limiting the utility of genetic data. This project’s sponsor, Dr. Toby Hocking, has created a machine learning algorithm that accurately detects and predicts actively transcribed regions of DNA sequences, known as peaks. The project’s objective is to make this machine learning algorithm accessible to gene scientists with no previous programming experience by integrating it into an easy-to-use web application known as PeakLearner. PeakLearner is integrated with Dr. Hocking’s machine learning algorithm and JBrowse, a genome visualization tool, in such a way that users are able to interact with the software’s peak prediction and detection analysis through a streamlined GUI. We have further increased the application’s ease of use by allowing users to log in with their Google Sign-In credentials and providing them the opportunity to work collaboratively. The successful implementation of the PeakLearner web application will provide gene scientists across the globe with a sophisticated and accessible tool for identifying areas of the genome that are actively expressed. There, regions are likely to be perturbed from a wild-type state in many pathological conditions, highlighting the current unmet need that PeakLearner will address.

If you have comments about this submission, email us here!