tree branch visibility is incorrect

[jameshadfield]

As it stands, any branch that is present within the selected time slice is "visible". E.g. the branch joining a tip from 2016 with a parent in 2013 will be selected if the max date is 2014.

Furthermore, as both tips & branches use the same visibility array, the 2016 tip (in the above example) is shown as visible.

This screenshot indicates zika filtered to a max date of 2014:

[trvrb]

I've been annoyed by this as well. My only thought for a solution has been to break up branches into multiple subbranches, each with 1 parent and 1 child node. However, I was worried about this for performance reasons as it would significantly increase the number of SVG objects, which has already shown to be a performance bottleneck.

Alternatively, there could be some way to change x-values for nodes dynamically based on time interval selected. This would give more of a growing tree effect than the current approach.

[jameshadfield]

You could do it with one extra SVG line per branch and using stroke-dasharray or changing the d attribute. With the way phylotree works, this isn't easy.

[trvrb]

Yes. I'm afraid of digging too deep into Phylotree and would instead be thinking of solutions that change the "data" supplied to phylotree.

[jameshadfield]

@evogytis and I have a possible solution to this:

• On the initial rendering of a tree (or after layout changes, zoom into clade etc etc) draw the skeleton tree - the entire tree with thin branches - what's currently the "not visible" state.
• When rendering "true" branches, only render those which are visible.
• If a branch is visible, but intersects the time slice, only draw the portion of the branch which is in view (i.e. modify the x co-ordinates)

This should actually make animation, filtering etc much faster, as the skeleton tree is left unchanged and so we are rendering far fewer branches at each time point.

[jameshadfield]

Closed in #639