Deep Learning Methods for Analysis of Genomic Data

Presented by: Zhi Wei, PhD, Professor, Department of Computer Science, Ying Wu College of Computing, New Jersey Institute of Technology

Thursday, September 24th
2:00pm – 3:30pm

In this workshop, two applications are used to illustrate how deep learning methods can help to solve biological problems. The first one concerns about clustering analysis of single-cell RNA-seq (scRNA-seq) data. Clustering is a critical step in cell-based scRNA-seq discovery studies. Conventional unsupervised clustering methods may not always produce biologically meaningful clusters with good interpretability. The user may have to manually and repeatedly tweak clustering parameters until acceptable clusters are found. We introduce a deep learning method that can integrate biological domain knowledge into the clustering step for producing more biologically meaningful clusters.

The second application concerns about copy number variation (CNV) analysis in genetics. CNVs are an important class of variations contributing to the pathogenesis of many disease phenotypes. Detecting CNVs from genomic data remains a difficult problem that even state-of-the-art methods suffer an unacceptably high false-positive rate. A common practice is to do manual filtering post CNV calling. Such a strategy is laborious and may introduce bias, which is not desired in a clinical setting. We introduce a deep learning method that helps to automate the filtering process. We show that the quality of CNV calling is much improved as a result.

To Join the workshop:


For questions and RSVP contact: