Event:

Systems biology aims at the model-based interpretation of biological data. The statistical and/or dynamical models are abstractions of biological processes, validated with the measured data, and used to guide further experimentation. Current methodological challenges arise from low sample numbers, noisy measurements and biological variability.
In this talk, I will show how we can account for these challenges in the inference process on different scales using probabilistic and Bayesian modeling. First, I will focus on parameter estimation in chemical reaction kinetics, which we describe as a combination of ordinary differential equations and mixture modeling. We use this ODE-constrained mixture modeling to study NGF-induced Erk1/2 phosphorylation in primary sensory neurons and reconstruct static and dynamical subpopulation characteristics across experimental conditions. I will then switch gears and review our large-scale approaches on genomic data sets observed on the single cell level, and how we quantify heterogeneities there. We deconvolve the observed heterogeneity by explicitly modeling latent causes of variation, which I will illustrate this on single-cell qPCR and RNAseq data from cellular differentiation decisions.