BUSTED Results as a Bipartite Network

This is me exploring bipartite networks again, this time w BUSTED results. These are the same data as the 'Support for postive selection' heatmap in the BUSTED results summary demo page. We're filtering ER at >10 and only showing codons w links. Thats just for readability, well have to sort something else for real use if we like this.

Columnar bipartite network

Ideas for real use: two filters one for pos ER, another neg. stroke width for pos and neg ER might be different.. should color by pos or neg too, just two hues like red + blue.

alt idea might be to have a gradient diverging colormap w midpoint at gray = 1, and rely on the very small stroke width and light color to help w readability ? that would eliminate the filters, but not the need for different stroke width scales for pos and neg ER..

Force directed bipartite network

same idea as the other.. different layout. codons are smaller circles, branches larger.

pros: pretty obvious when a branch doesnt have much going on (bc its off in space), or is doing something different to other branches (bc its connected but still kinda off in space lol).

cons: overlaying phylogeny info meaningfully is a bit of a difficulty.. i think for realz nodes wouldnt get colored by group, bc we need to color links.. but they might be differentiated by shape or some other form of highlighting upon user-selected clades? maybe there would be an independently scrolling phylotree to left that a user could choose clades from? also, the only real opportunity for codon info here is in tooltips or similar.

my 2c, i dont currently think this is a primary result summary viz, but could be a complimentary, more exploratory one. it does seem fun though, something offered after we get the primary thing sorted.