Abstract
We use molecular dynamics simulations based on publicly available MNase-seq data for nucleosome positions to predict the 3-D structure of chromatin in the yeast genome. Our main aim is to shed light on the mechanism underlying the formation of micro-domains, chromosome regions of around 0.5-10 kbp which show enriched self-interactions, which were experimentally observed in recent MicroC experiments. We show that the sole input of nucleosome positioning data is already sufficient to determine the patterns of chromatin interactions and domain boundaries seen experimentally to a high degree of accuracy. Since the nucleosome spacing so strongly affects the larger-scale domain structure, we next examine the genome-wide linker-length distribution in more detail, finding that it is highly irregular, and varies in different genomic regions such as gene bodies, promoters, and active and inactive genes. Finally we use our simple simulation model to characterise in more detail how irregular nucleosome spacing may affect local chromatin structure.