Applied Mathematics Colloquium – Zeliha Kilic, Department of Physics, Arizona State University

Dr. Zeliha Kilic, Postdoctoral Research Associate in Pressé Lab

Department of Physics, Arizona State University

http://zkilic.wixsite.com/zelihakilic 

Talk taking place over Zoom.

Title: BAYESIAN APPROACHES to DATA INTERPRETATION IN APPLIED PHYSICAL SCIENCES

Abstract: Fluorescence experiments based on Förster resonance energy transfer (FRET) present a powerful way to probe switching kinetics between the conformational states of  a molecule. The acquired data from these experiments generated by a single molecule are traditionally interpreted with hidden Markov models (HMMs). The underlying modeling assumptions for HMM framework is that the switching kinetics of the molecule has to be slower compared to the data acquisition rate. The problem is, single molecule kinetics are unknown in principle. Therefore, it is unclear, a priori, whether HMM framework applies to the probed single molecule kinetics. In this talk, I will present an overview of what we propose for the generalization of HMMs for single molecules as they switch between their conformational states. I will demonstrate a variant of HMM generalization performance on experimental data from FRET labeled Holliday junctions. I will also highlight the novel Bayesian non-parametric approaches on how to tackle with the model selection problem when the conformational states of the molecule are unknown.