Title: Particle tracking in Biology using machine learning
Speaker: Jay Newby, University of Alberta; jnewby@ualberta.ca
Date: Sunday, June 13, 2021
Time: 13:00 - 17:00 (EDT)
Abstract:
There are two basic ingredients for particle tracking:
- microscopy videos of nanometer to micrometer sized "particles" suspended in a fluid and
- a stochastic model of particle motion.
Given these two ingredients, we can use machine learning methods to gain insight into micron-scale systems. Particle tracking has many applications in physics, chemistry, and biology. We will be focusing primarily on the latter. Some examples of "particles" are synthetic beads, genetically expressed fluorescent proteins, biopolymers, viruses, and bacteria. The motion of small particles in a fluid is a stochastic process. The classical example is Brownian motion, which was originally discovered through observation of pollen suspended in water. Once microscopy videos are obtained, the position of each particle is tracked through time. The result is a set of position-time series tracks. The tracks are typically used to infer properties of the fluid. The first example discovered was through observation of Brownian motion. The simplest stochastic model of Brownian motion involves a single parameter, the diffusivity. The Stokes-Einstein relation is a formula that relates the diffusivity and particle size to the fluid viscosity and temperature. In particle tracking microrheology, particle motion is used to estimate the viscosity and elastic properties of a non Newtonian visco-elastic fluid. In biology, many new applications for particle tracking are beginning to emerge, thanks to advances in microscopy, machine learning, and neural networks. A few examples are characterizing active bacterial motion of Salmonella in mucus and measuring macromolecular crowding in the cytoplasm of living cells.
Requirements:
This workshop assumes a basic understanding of probability and stochastic processes. Some amount of programming will be involved in all of the projects (students with complementary skills will be encouraged to form teams). We will primarily be using Python (but R, Julia, or C++ might be ok too). Students will need to bring a laptop or tablet equipped with a keyboard. The only required software is the Google Chrome internet browser with a Gmail or other Google account logged in.