Computational 3D genomics
The focus of the group is to understand how the three-dimensional (3D) organization of DNA inside the nucleus relates to critical functions in the eukaryotic cell, such as gene expression regulation and the cell cycle. We use computational modeling, statistical analyses and machine learning to elucidate evolutionary trajectories and functional relationships across cell-types, tissues and species.
A central technology in the group is the Hi-C technique. We build computational tools to analyze these and related data, to increase our understanding of the eukaryotic genome. Below, are some recent (and ongoing) projects.
Chrom3D – Three-dimensional genome modeling from Hi-C data
Chrom3D is our genome 3D modeling platform – designed to integrate Hi-C data with other positional constraints based on association of loci with intranuclear anchors. The combination of Hi-C and other information enables genome-wide radial positioning of TADs in ensembles of 3D structures. In several ongoing projects, Chrom3D is consistently maintained and developed further. Chrom3D is available on Github: https://github.com/Chrom3D/Chrom3D
– Paulsen et al. (2017). Chrom3D: three-dimensional genome modeling from Hi-C and nuclear lamin-genome contacts. Genome Biology
– Paulsen et al. (2018). Computational 3D modeling using Chrom3D. Nature Protocols.
Statistical models of genome contact frequency maps
We are developing statistical models of contact frequency maps from single-cell (and ensemble) Hi-C, ChIA-PET and related data. These statistical models allow us to explore the data in new ways, to learn about the underlying properties of genome organization. For example, we have determined a set of intrinsic structural principles giving rise to correlations between distal genomic regions. These need to be accounted for during inferential analysis from these contact maps. Further, we have developed models that allow us to directly infer 3D structural models of chromosomes from single-cell Hi-C data.
– Paulsen et al. (2015) Manifold based optimization for single-cell 3D genome reconstruction. PLOS Computational Biology
– Paulsen et al. (2013) Handling realistic assumptions in hypothesis testing of 3D co-localization of genomic elements. Nucleic Acids Research
– Sekelja et al. (2016) 4D nucleomes in single cells: what can computational modeling reveal about spatial chromatin conformation? Genome Biology
Genome domains: understanding their functional and evolutionary basis
Eukaryotic genomes are non-randomly organized. Even though the mechanisms underlying genome structure regulation can vary across species, some general “rules” related to gene expression differences seem to apply. Most species organize their chromosomes into separate territories within the nucleus, and within the chromosomes different regions are usually organized into compartments and/or domains. In mammals, Topologically Associated Domains (TADs) is a prominent feature seen along the diagonal of the Hi-C contact matrices (see images above). Many of these TADs seem to be formed by a proposed process termed Loop extrusion. Together with Philippe Collas, we have identified a subset of TADs (TAD-cliques) behaving similar to compartments, yet with distinct features. These TADs are altered during differentiation in concert with formation of heterochromatin compartments in the nuclear periphery.
– Paulsen et al. (2019) Long-range interactions between topologically associating domains shape the four-dimensional genome during differentiation. Nature Genetics
We are interested in a range of other genomics topics as well, and are always interested in collaborating with groups that work with these (and related) topics:
- Genome assembly. Especially using Hi-C assisted approaches
- Epigenetics and gene regulation
- Somatic mutations in cancer and other diseases
- We can also help with development of bioinformatics tools/software
4D nucleome modeling
Curr Opin Genet Dev, 67, 25-32
Long-range interactions between topologically associating domains shape the four-dimensional genome during differentiation
Nat Genet, 51 (5), 835-843
Lamin A, Chromatin and FPLD2: Not Just a Peripheral Ménage-à-Trois
Front Cell Dev Biol, 6, 73
Computational 3D genome modeling using Chrom3D
Nat Protoc, 13 (5), 1137-1152
The lipodystrophic hotspot lamin A p.R482W mutation deregulates the mesodermal inducer T/Brachyury and early vascular differentiation gene networks
Hum Mol Genet, 27 (8), 1447-1459
Carcinogen susceptibility is regulated by genome architecture and predicts cancer mutagenesis
EMBO J, 36 (19), 2829-2843
Chrom3D: three-dimensional genome modeling from Hi-C and nuclear lamin-genome contacts
Genome Biol, 18 (1), 21
Hi-C-constrained physical models of human chromosomes recover functionally-related properties of genome organization
Sci Rep, 6, 35985
4D nucleomes in single cells: what can computational modeling reveal about spatial chromatin conformation?
Genome Biol, 17, 54
Galaxy Portal: interacting with the galaxy platform through mobile devices
Bioinformatics, 32 (11), 1743-5
Manifold Based Optimization for Single-Cell 3D Genome Reconstruction
PLoS Comput Biol, 11 (8), e1004396
The Genomic HyperBrowser: an analysis web server for genome-scale data
Nucleic Acids Res, 41 (Web Server issue), W133-41
The genome sequence of Atlantic cod reveals a unique immune system
Nature, 477 (7363), 207-10