Coloring by nt shows non-N bases, but amino acid is showing X where unmutated

[AngieHinrichs]

Hi Theo! Viewing a tree from @iqbal-lab, when I choose Color by: Genotype and it defaults to S:484, the unmutated parts of the tree are all grey and mouseover shows 'X' for the grey sequences:

It's even more striking for a position like 377 - almost the entire tree is grey. But if I color by nucleotide at the same positions (22691-22693, thanks for codon2nucleotide 😃) then all nucleotide bases are colored correctly from the root of the tree.

[theosanderson]

Hey so unfortunately when I overhauled GenBank import in taxoniumtools a few months back to deal with compound features I broke the annotation of the root node without noticing. This was recently fixed https://github.com/theosanderson/taxonium/releases/tag/v2.0.111 but probably too late for this tree. Really sorry for the inconvenience. Please let me know if it persists in any new trees.

[AngieHinrichs]

Ah, thanks Theo, so this should just go away when they update taxoniumtools and re-run. I'll close this now and reopen if that doesn't do the trick.

[martinghunt]

Thanks Theo, can confirm that the tree was made using taxoniumtools 2.0.110. If I rerun with 2.0.111 then it's mostly fixed.

But I am getting a glitchy display issue where depending on zoom level everything turns grey. Maybe related? I don't know. I've not seen this happen before.

Two screenshots attached of the same session, only difference between them is I used the mouse wheel to zoom in/out. Running on taxonium.org/M1 mac/safari.

[theosanderson]

@martinghunt - huh! that's a new one to me.

Some questions:

1. Can you get this to happen at cov2tree.org?
2. I've made some changes to some bits of (web) Taxonium lately to add new colour customisability. https://cov2tree-9k82lpuou-theosanderson.vercel.app/ is a version before those changes - does that have the same bug? If not it might be a temporary fallback for you
3. Is this generally very reproducible, i.e. does it happen every time?

If it doesn't happen on cov2tree.org please could you send a file where it happens over to theo@theo.io - dropboxed or whatever?

[theosanderson]

OK I can reproduce this - doesn't happen with web taxonium but does with a JSONL

[theosanderson]

It indeed appears that https://cov2tree-9k82lpuou-theosanderson.vercel.app/ is unaffected if that helps in the short term

[martinghunt]

Thank you @theosanderson, yes https://cov2tree-9k82lpuou-theosanderson.vercel.app/ is working perfectly for me. Will use that for now.

[theosanderson]

Just using this ticket for notes to myself for a bit:

• easiest way to reproduce is zoom right in when not on "genotype" mode, then switch to genotype mode. works for cov2tree too.
• relates to the colorCache in useColorBy - if you disable caching it stops happening
• presumably there is an initial point where it can't find parents and so cache gets set to "X". that initial point where it can't find parents shouldn't happen.

[theosanderson]

This PR, is the issue.

https://github.com/theosanderson/taxonium/pull/519

We can walk through it to see when the issue appeared. 2204d84 bad e8bbbc4 good

[theosanderson]

Hopefully a fix is rolling out, preview at https://cov2tree-git-fix-genotype-theosanderson.vercel.app/ . If it doesn't fix the issue please let me know. Thanks for reporting

[martinghunt]

Thanks for fixing so fast! That preview is working perfectly for me

[theosanderson]

Awesome - (should be same on main site)