Quantifying differences between related proteins
||Thursday, November 3, 2016
||12:00pm - 1:00pm
||D344 LSRC, Duke
Most human proteins are part of large structural families. Although closely related proteins have similar structures, they oftentimes play different roles in the cell. We focus on a special class of proteins called transcription factors (TFs), which bind to short DNA sites across the genome and regulate processes such as cell growth and differentiation. TFs that belong to the same protein family are currently believe to have identical DNA binding specificities, as existing specificity models are simply indistinguishable among family members. Based on high-throughput data generated in our lab, we develop and use different regression approaches (stratified SVR models, stable LASSO, weighted least square regression, Dirichlet process mixture models) to detect and quantify differences between TFs currently believed to bind identical DNA patterns. We show that such differences help us explain the TFs' different roles in the cell, and that they are relevant for interpreting the impact of DNA mutations in our genomes.
Raluca Gordan is an assistant professor at Duke University, with primary appointments in the departments of Computer Science and Biostatistics and Bioinformatics. Her primary research interests are in computational regulatory genomics. She develops computational methods, as well as high-throughput experimental techniques, to quantitatively characterize protein-DNA binding events. Previously, she was a postdoctoral fellow at Harvard Medical School (2009-2011) and a PhD student in Computer Science at Duke University (2005-2009). Her recent awards include a Sloan Fellowship (2014), a Basil O'Connor Award from the March of Dimes Foundation (2013), and a Research Starter Grant in Informatics Award from PhRMA Foundation (2013).