Back to Colloquia
Physics Colloquium - Thursday, Nov. 6th, 2008, 2:30 P.M.
E300 Math/Science Center; Refreshments at 2:00 P.M. in Room E200
Ilya Nemenman - Information Sciences Group (CCS-3)Los Alamos National LaboratoryInformation theoretic methods biology: from information in spikes to predicting protein properties
Modern biological data analysis often relies on estimation of information-theoretic (entropic) quantities from empirical data to study problems as diverse as the structure of the neural code, the reverse-engineering of biochemical networks from high-throughout abundance data, such as gene expression microarrays, and estimation of binding free energies of macromolecules. Unfortunately, information/entropy estimation suffers strongly from insufficient sample size and the consequent bias, which has limited their utility so far. However, in the course of the last five years, we have developed a series of methods that alleviate the bias problem, often requiring a square-root-fewer samples than traditional methods to achieve the same level of accuracy. In this talk, I will describe the methods and will illustrate their broad applicability by focusing, in particular, on two biological problems: understanding submillisecond precision in neural responses of the blowfly H1 neuron and predicting lattice polymer configurational entropies.