Notes

Other viewers:

• https://bsiranosian.shinyapps.io/shiny_structure_explore/
• http://3dgenomes.github.io/TADkit/#!/project/browser?conf=assets%2Fexamples%2Fconf.json
• http://delta.big.ac.cn/
• http://csynth.org/

Data:

• https://github.com/BogdanBintu/ChromatinImaging
• http://sgt.cnag.cat/3dg/datasets/files/20180717_Nir_etal_2018_Dataset.zip

Online viewer for the Nir et al models (pdb).

• http://jenalib.leibniz-fli.de/cgi-bin/3d_mapping.pl?CODE=

Note: Data is in "data/Nir/Models". The Nir PDB model files are broken, there is no space between the x&y coordinates. These are easy to fix, find the position which needs a space and use sed. For example, file CS1A.pdb was fixed as follows

sed 's/.\{38\}/& /' CS1A.pdb

IGV session for IMR-90 CTCF and RAD21 tracks:

• http://bit.ly/2t4iDrA

Noisy data

The data is noisy, and there are occasional crazy outliers. What to do about it?

• Have option to right-click and remove a sphere. The structure would then be broken there (same as a NAN)
• Automatically remove spheres > certain distance from their neighbors. We know the genomic distance between centroids, so we can estimate with good accuracy the maximum possible physical distance. The distance between base pairs in a double helix is ~ 3.4 Angstroms (0.34 nanometers). So in a 30kb window the centroids cannot be more than ~10,000 nm.