Computational Methods for Predicting Changes in 3-D Chromosome Re- arrangement and Gene Deregulation in Human Diseases

The aim of this project is to develop computational models in order to predict the 3-D chromatin organization. This could be achieved by starting from the HiP-HoP model which allowed to reproduce the conformation of the Pax6 locus obtained through 3C and FISH experiments.
The basic idea of HiP-HoP simulations is to represent the chromatin fibre as a bead- spring polymer where each bead corresponds to 1 kb. To understand how genomic loci fold, it is necessary to include proteins (i.e. transcription factors) which can bind particular sites along the chromatin chain forming molecular bridges. The polymer is so composed by beads of different types having a stronger or weaker interaction with transcription factors. Data from the ENCODE project allow to assign a specific type to each bead as they give information about epigenetic marks (e.g. H3K27ac regions), but also about chromatin accessibility (e.g. by using ATAC-seq).
This model embeds also possible post-translational modifications or active protein degradation by considering transcription factors as beads which can switch back and forth between a binding and a non-binding state.
Chromatin fibre folding is influenced by CTCF/cohesin loops too: to include this loop-extrusion (LE) mechanism, additional springs between non-adjacent beads are introduced in the HiP-HoP model. The springs move during simulations forming loops, and they can bind and unbind to the filament. This last mechanism intends to reproduce interactions between cohesin and CTCF binding sites, since the LE process driven by cohesin stops when this encounters a CTCF site with a motif oriented towards the direction of the extruder.
Lastly it is important to depict chromatin as a heteromorphic fibre in order to accu- rately predict the 3-D loci folding within cells with different levels of transcriptional activity. The heteromorphic chromatin is obtained by including additional springs between beads without acetylation mark, leaving H3K27ac regions less compact. All previous features characterize the HiP-HoP model which correctly reproduces the folding of the Pax6 locus, but also of the globin loci, without requiring any fitting to experimental data (e.g. HiC data). The aim of this PhD project is therefore to start from this model to predict confor- mations of loci in other cell types and different organisms.