I’m currently a 4th year Ph.D. candidate in computational biology in the Erill Lab at UMBC.


Prokaryotic transcriptional regulation, biophysics, machine learning

Research Summary

Cells are chemical computers that accept environmental signals as inputs and express genes as outputs. An even shorter way of putting this is just to say that cells compute their metabolism. I’m interested in helping to understand this process through applications of information theory, machine learning and statistical physics to the study of transcriptional regulation in bacteria.

Optimal Recovery of Binding Energy Models from ChIP-Seq

Transcription factors (TFs) are a major substrate of cellular information processing, connecting environmental and intracellular signals to responses in gene expression. As a consequence, an overwhelming amount of research in molecular biology is somehow connected to their study. While a typical bacterial species might possess over two hundred distinct TFs, most traditional techniques for characterizing them are laborious and expensive. Developing new methods to maximize the information gain from biological experiments and automating the discovery of transcriptional regulatory elements from sequence data are therefore pressing problems in computational biology.

At the same time, progress in genome sequencing and chromatin immunoprecipitation (ChIP) techniques has placed molecular biology at the forefront of the ‘big data’ revolution in scientific practice. Although sequence data is currently accumulating at a rate that outstrips Moore’s law, our ability to extract meaning from this deluge is limited. Currently, most ChIP experiments are analyzed in a heuristic, semi-quantitative fashion, but the resolution of the resulting datasets suggests that this limitation is largely self-imposed.

I’m interested in developing new methods for the automatic recovery of higher-order features of transcriptional regulatory systems, such as multimerization and DNA looping.

In-Silico Evolution of Transcriptional Regulation

I’m also interested in characterizing interactions between transcription factors and their cognate binding sites. For a given functional specification there are often many “programs” which can encode roughly equivalent behaviors– how does nature decide? From a collection of binding sites known to be co-regulated by a transcription factor, what can be inferred about the system requirements of the network it participates in and the evolutionary dynamics which produced it? These questions lead us to develop mathematical and computational models of protein-DNA co-evolution in order to understand the design constraints that gave transcriptional regulatory networks their present wiring.