Nava, A. (AM) – Machine-Learning Methods for Prediction of Biological Systems

Advances in microscopy have enabled the collection of high-quality single-cell datasets, providing new opportunities to identify the mechanisms underlying complex biological processes. In this work, we develop machine-learning frameworks using single-cell temporal data with the goal of predicting and providing insights into these mechanisms. We produce frameworks for two biological systems, bacterial spore germination, the process in which bacteria begin metabolic activity, and embryonic stem cell organization. Bacterial spore germination is a critical transition in which the spore becomes susceptible to control techniques, however the mechanisms governing this transition are unknown. We develop a machine-learning framework that predicts germination timing at the single-spore level, enabling identification of predictive features associated with germination that may reflect underlying biological mechanisms. Embryonic stem cell organization has been shown to closely recapitulate formations seen in embryonic development, but the mechanisms driving their spatial organization remain unclear. Here, we develop a simple agent-based model in which spatial organization is driven by cell-cell interaction parameters. We then train a machine-learning framework to infer these underlying interaction parameters from simulated data and propose that this approach can be extended to other agent-based models calibrated to experimental stem cell data. These studies demonstrate that machine-learning models can be used for prediction using single-cell temporal data, as well as tools to develop mechanistic hypotheses.

Event Host: Alexandra Nava, Ph.D. Student, Applied Mathematics 

Advisor: Marcella Gomez

Zoom: https://ucsc.zoom.us/j/98821445104?pwd=OFAKwGrObh02bPLgXieXsDcTxxS1Cj.1

Passcode: 392769