Biological questions to be addressed (From JAWG):
How does a loop, TAD, Compartment (From Hi-C definition) look like?
How often do these domains appear on a per single cell basis? Are these imaging results compatible with corresponding Hi-C data based on populations? Is there heterogeneity from cell to cell?
What are the dynamic properties of these domains? Can we make predictions?
Based on these questions, the Imaging Approaches to Genomics (IAG) will work to define an experimental strategy to generate imaging data that can be correlated to biochemical/sequencing data. In addition, even after image collection we need a common image data analysis format to correlate imaging and omics data (IWG current work).
The IAG will address the following points, in order:
Determine limitations of labelling approaches: DNA-FISH, CAS-labelling, unspecific labelling (e.g. replication domains). How many discernible labels in a single experiment? What is the smallest unit of DNA that can be realizably labelled and visualized with either DNA-FISH or with CAS-labelling?
Determine the spatial and temporal resolution limits of microscopy technique(s): live-cell microscopy, super-resolution microscopy and potentially electron microscopy.
Prioritize labelling approaches and microscopy techniques to image loops, TADS and domains, respectively. Each genome organizational layer happens at different spatial scales, is likely going to be best imaged with different microscopy techniques, and likely poses different technical challenges.
Prioritize loci to image (From original JAWG list). At least 2 loci for each level (loop, TAD, compartments). Possibly choose loci for which there are already available genetic mutants (CRISPR), and/or whose conformation predictably changes in response to a biological stimulus.
Prioritize imaging throughput/dynamics and image analysis choices. How many loci do we need to image for experimental condition (Heterogeneity)? How fast can we image loci in live cells (dynamics)? What kind of data do the omics bioinformaticians want: position, distances, movement (IWG ongoing work)? All three? Is there a fundamental limit in correlating structures from sequencing and imaging?