Circle plots to show pairwise quantities for domains, chains, ...

[olibclarke@…]

Begin forwarded message:

From: Oliver Clarke
Subject: ChimeraX suggestion: chord diagrams (circos plots) to visualize protein-protein or domain-domain interactions?
Date: February 17, 2016 at 3:47:43 PM PST
To: Tom Goddard

Hi Tom, another one, take or leave:

I think one thing that we're missing in Coot/Chimera/Pymol etc is a really clean way to represent subunit-subunit or domain-domain interactions, in a way that takes account of distance in sequence space and allows for easy representation (and potentially selection) of both intra- and inter- chain interactions. Distance matrices are one way, but they can be a little confusing and difficult to interpret at first glance, and don't scale well to analysis of multiple chains (or protein-ligand interactions for example).

In genomics, a lot of people have been turning to the use of chord diagrams (also called circos plots) for the analysis and visualization of exactly this kind of pairwise interaction data (see ​http://circos.ca/software/ , ​http://www.delimited.io/blog/2014/11/18/interactive-chord-diagrams-in-d3 (I really like the interactive approach but not sure whether it is possible outside of D3.js), and R documentation of circlize or Rcircos for examples - I believe for python ggbio in Bioconductor may be used for this kind of diagram).

I wonder whether a similar kind of plot might be helpful for visualizing either distances, or changes in pairwise distances between conformations, in structural analysis - e.g. map the sequences of all chains in the structure around the outside of a circle as boxes, and show all pairwise interactions that meet criteria (greater than x distance in sequence space, closer than y Å) as chords across the circle. I am going to have a go at doing this with one of my structures if I can figure out how to get the data properly formatted - of course, I am not sure how much you guys want to get into this kind of analysis in ChimeraX, it is just a thought.

Cheers,
Oli.