Detecting T Cell Activation Using a Varying Dimension Bayesian Model

J Appl Stat. 2018;45(4):697-713. doi: 10.1080/02664763.2017.1290789. Epub 2017 Feb 16.


The detection of T cell activation is critical in many immunological assays. However, detecting T cell activation in live tissues remains a challenge due to highly noisy data. We developed a Bayesian probabilistic model to identify T cell activation based on calcium flux, a dramatic increase in intracellular calcium concentration that occurs during T cell activation. Because a T cell has unknown number of flux events, the implementation of posterior inference requires trans-dimensional posterior simulation. The model is able to detect calcium flux events at the single cell level from simulated data, as well as from noisy biological data.

Keywords: Bayesian; Indo-1; MCMC; Pseudo prior; T cell activation; Two photon microscopy.