Skip to ContentSkip to Navigation
Research Bernoulli Institute Calendar

Danilo Pellin, MSc: Stochastic modelling of dynamical systems in biology

When:Mo 03-04-2017 10:00 - 11:00
Where:5161.0222 (Bernoulliborg)

In this talk two relevant biological problems will be addressed from a statistical modelling perspective. The first regards the study of hematopoiesis, a still not well understood biological process rarely observable in humans due to technical and ethical reasons. Hematopoiesis is responsible for the production and replenishment of all blood cellular components and occurs throughout life. Improving our knowledge about the dynamics of this process can help scientists and clinicians to better understand the mechanisms behind blood disorders and cancers, refine transplantation protocols, define novel strategies for diseases with currently unmet clinical needs.

Exploiting follow-up data collected from three patients recruited in a gene therapy clinical trial for Wiskott-Aldrich syndrome, a rare blood disorder, we are able to track the in vivo the evolution of several corrected hematopoietic stem cells over multiple lineages. Modelling individual clone dynamics by means of a continuous-time Markov process, we derive systems of ordinary differential equations for process moments’ dynamics. Based on these, we develop method-of-moments procedures to infer both the set of lineage-specific rates governing the stem cells differentiation process and the hematopoietic tree structure.

The second biological investigation lies in the context of neuroscience. We focus our attention on neuron communication at the level of its most elementary operating systems. We propose a new comprehensive generative model for spontaneous neurotransmitter release occurring at single synapses, based on the superimposition of renewal processes. Our approach is able to mimic the characteristic presence of bursting episodes followed by long intervals of synaptic inactivity. The estimation procedure is based on both Monte Carlo simulation and a quantile-based scoring function. The analysis of experimental data sets partially confirms previous estimates for the parameters governing this complex biological process.