Computational 3D genome modeling
Annaël Brunet, Tharvesh M. Liyakat Ali
Computational methods to infer 3D structural views of the genome in the nuclear space can reveal spatial relationships between genomic regions that are not visible in the underlying data. Physics-based modeling infers genome folding principles based on physical properties of chromatin. Other genome-modeling approaches consider interaction probabilities between genomic domains determined from e.g. Hi-C data, to infer a 3D positioning of these domains. Analysis of the models provides insights into 3D genome conformation. Spatial genome investigations can also generate hypotheses testable in a wet-lab. For a recent review from our lab, see Sekelja et al. 2016 Genome Biol.
We are developing computational methods and tools for 3D structural modeling of the genome at multiple scales and using interaction probabilities and polymer physics approaches.
- Computational methods for 3D and 4D modeling of genome architecture
- Relationships between 3D chromatin folding, nuclear architecture, epigenetic states and gene expression potential in time-series
- Chrom3D genome structures in Virtual Reality (see our video on Youtube)
- Chrom3D: a computational platform for 3D genome modeling from HiC and lamin-genome contacts (Paulsen et al 2017 Genome Biol 18:21)
- ChIP protocol for nuclear lamins (Oldenburg et al 2016 Meth Mol Biol 1411, 315-324)
- manifold-based optimization to enhance modeling of 3D chromatin structure from sparse HiC data (Paulsen et al 2015, PLoS Comput Biol 11, e1004396)
- EDD: a domain calling algorithm (Enriched Domain Detector; https://github.com/CollasLab/edd) to map LADs and other broad domains from ChIP-seq data (Lund et al 2014 Nucl Acids Res 42, e92)