Computational Methods for Integrating Microscopy with Chromatin Structures
The genome is more than a linear sequence of bases; its spatial organisation is a key part of its function. In humans, three billion base pairs, or approximately two metres of DNA are packaged into a nucleus a few micrometres in diameter. The genome must also be organised so that it can be replicated and partitioned into daughter cells, and so that regulatory elements are positioned to affect their targets.
Until recently, little was known about the organisation of the genome at the scale of single genes. The packaging of DNA onto nucleosomes, and the segregation of chromosomes into chromosome territories was well understood, but the development of chromatin conformation capture (3C) techniques has enabled the first thorough study of intermediate scales. These methods provide information about the distances between pairs of genomic loci, which gives indirect information about their positions. By applying these techniques to single cells, it has become possible to calculate a structure from the observed distance restraints. Through the prior constraints placed on the model, such as the existence of a continuous backbone, these structures provide additional information about the conformation of DNA.
To overcome the limitations of 3C, it is useful to integrate additional sources of information. I present several methods for the validation and improvement of Hi-C structures by adding data from microscopy, and for characterising dyes used in single-molecule light microscopy. It is found that single-cell Hi-C structures agree with fluorescence microscopy when observing the distance of genes from the edge of the nucleus, and that centromeres are not a suitable label for direct validation. Adding absolute positional restraints from images is shown to be useful in better determining chromatin structure in synthetic tests. Finally, the presence of a FRET acceptor near a fluorescent protein is shown to improve its photophysical properties.