丁香园AV

*To request access to the videoconference, email dsivak@sfu.ca

Synopsis

By what mechanism do clusters of transcription factors鈥攊n constant and rapid protein subunit exchange with their environment鈥攈ierarchically assemble and regulate transcription? How do clusters, such as clusters of T-cell receptors or estrogen receptors, differ in assembly kinetics and composition across cell membranes? Can we follow the reaction kinetics of interacting pairs of biomolecules in an effort to probe their heterogeneity, such as the human linker histone H1 and histone chaperone prothymosin 伪?

Such questions recently posed with state-of-the-art imaging tools present profound data analysis challenges for these fundamental reasons: 1) they involve many-protein processes resolved at the single molecule (SM) level often indirectly monitored through light (by fluorescence); 2) these processes occur on fast imaging timescales (micro to tens of ms); 3) events may involve as many as tens or more molecular actors all localized to regions below light鈥檚 diffraction limit.

Deriving insight from such experiments means disentangling complex, fast and highly stochastic SM dynamics鈥攖hat we care about鈥攆rom the equally stochastic measurement output to which the SMs are inexorably tied. To obtain models of SM dynamics on timescales approaching those of data acquisition and on length scales below light鈥檚 diffraction limit demands novel Mathematics that go beyond the reach of those currently employed. Here we discuss our role in bringing to light new tools from Data Science and Computational Statistics鈥擝ayesian nonparametrics (BNPs)鈥攖o the Natural Sciences and discuss their unique role in resolving the most pressing challenges at the heart of SM problems.