CAREER: Computational Methods for Learning Dynamic Networks of Biological Regulation and Control

Duke University, Durham NC

Investigators

Abstract

The objective of this research program is twofold: first, to develop and apply new computational inference methods for elucidating the underlying mechanisms of complex biological systems, and second, to develop and enhance a series of courses that incorporate various aspects of computational biology into both the undergraduate and graduate curricula. Network inference methods based on dynamic Bayesian networks are being implemented and extended to discover models of complex biological systems. In addition, simulation systems capable of generating biologically plausible data are being developed to evaluate and improve the inference methods. The aim is to apply the most effective methods to real experimental data to produce models of the underlying biological systems. The inference methods being developed are widely applicable to biological and non-biological problems where models of complex systems need to be recovered from noisy or incomplete data. Both the inference algorithms and the simulators are designed to be flexible and adaptable to other systems. Insights gleaned during the course of this research program are being folded into new courses on machine learning and the dynamics of complex biological systems. These courses are intended to accompany the three existing computational biology courses previously developed by the PI. Collectively, these five courses will contribute to shaping both undergraduate and graduate programs in computational biology at Duke University.