Math Biology Seminar: Alejandra Herrera

STORM is a super-resolution technique that uses photoswitchable fluorophores to achieve resolutions at or below 20nm. A downside of STORM is the possibility of recording several blinks from one fluorophore, affecting the estimation of the number of molecules detected in the image. I constructed a mathematical model to identify unique fluorophores in STORM images by independently using the localization and the time series of the observations. The temporal sequence is described with a Markov chain approach and their spatial distribution with a Gaussian mixture model. To estimate the parameter values, I implemented a maximum likelihood procedure which requires a mixed optimization. Initially, I solved the mixed optimization problem with an extensive search algorithm on integers and a continuous optimizer for the rest of the parameters. I am currently investigating MCMC and Bayesian methods to speed up the optimization. I have tested my protocol in simulated data and I will use it to improve STORM images of B-cell surface receptors. B-cell receptors distribution on the membrane has been related to B-cell activation. This model will enhance a microscopy technique that is already widely used in biological applications and will allow to deeper analyze immune cells signaling.