Predictive modeling of genome structure

Posted on Posted in Research Projects

Mathematical models allow us to develop rich predictions from quantitative measurement.  This approach to date has been vastly more successful in physics than molecular biology, where models have enable us to predict the existence of fundamental particles long before they are detected in nature, or determine the mass of celestial bodies that we could never put on a scale and measure in the lab. Part of this distinction has been due to the challenges of getting reliable, quantitative data from inside cells – but thanks to new technologies, this is changing rapidly.  Building on the potentially large volumes of quantitative data we can extract by imaging and sequencing approaches, the Boettiger lab aims to use mathematical models to find simple explanations of the chromatin structures we observe and make new predictions about how these structures form or will be perturbed.  Here I describe one recent example of our work on the physical organization of Polycomb bound regions.

 

Case study: Organization of Polycomb bound DNA

Super-resolution imaging of Polycomb bound regions of the Drosophila genome revealed that these domains are densely packed and that longer regions are proportionally denser, such that the radius of gyration of a domain scales as ~1/5th power of the domain length.  Internally, the domain is not folded in a regular structure as commonly depicted for heterochromatin.  Instead, the domain folds back on itself in a structure reminiscent of an equilibrium globule, where many portions of the chain quite distantly separated in terms of the linear sequence are in close proximity in 3D.  Finally these domains show little overlap with non-Polycomb bound chromatin.

The unique packaging of Polycomb bound domains in terms of their dense organization, length-dependent density, high degree of intermixing and exclusion of neighboring domains can all be quantitatively reproduced with a simple polymer model in which weak, reversible homotypic interactions between Ph proteins bound to the domain form bridges between different parts of the polymer backbone.

3d-genome-modeling