Closed emmahodcroft closed 5 years ago
Thanks @emmahodcroft -- let's discuss in the UK?
Here are some quick notes while I have internet:
We need a way to specify this in augur export V2 - not obvious best idea?
I can't think of any way apart from a config file unfortunately. We could also provide command line args, which may be enough for simple cases.
clade_annotation
isn't displayed as text labels on nodes on the tree
in v2 these aren't traits, they're under the labels
property on nodes, so we'll have to change augur export to do this.
Nucleotide colorby doesn't seem to be working properly?
:scream:
P.S. Great image captions.
P.P.S. Thanks so much for actually testing out auspice v2 -- it's immensely helpful
@emmahodcroft
Sorry, working through all this in an order that may not match how they came in 🙃 . Most of this I address in latest comment on issue #5!
Yes - clades is fixed!
"Traits/Geo" - I didn't realise it above but this is somewhat two issues: if you specify something as --geography-traits
it needs to be written on the nodes in export JSON to work, but does not necessarily need to be a colorby (if we so decide). Currently these two things are rather entangled, but we should separate them out better - see my comments on the other issue.
For excluding trait colorbys and also about 'filters', see the other issue! :)
cool. Closing this as I think everything's being talked about in other issues. Re-open if not!
Traits
Issue: V1 JSONS allows you to specify
legendTitle
&menuItem
(fromconfig.json
, not used in V2), V2 does not. V2 just uses whatever thekey
/traitvariable is called (ex:age_range1
instead of 'Age Range 1') Where?augur export
for V2 JSONS.title
field is the same as thekey
/traitvariable. Missing feature or bug? Missing feature. We don't have any way (AFAK) to say what thetitle
should be. If you manually replace thetitle
field incolorings
in V2 JSONS, it displays correctly. Also displays correctly for V2 JSONS generated from V1 JSONS (withauspice convert
) Solution? We need a way to specify this inaugur export
V2 - not obvious best idea?Issue: V1 JSONS viewed in Auspice 2.0 use
legendTitle
rather thanmenuItem
for the drop-down colorby menu. If they aren't the same, can lead to uninformative menu items. V2 JSONS generated from V1 JSONS (withauspice convert
) also have this problem. (ex: For mymenuItem
s 'Age Range 1', 'Age Range 2', and 'Age Range 3', all use just 'Age Range' aslegendTitle
. In Auspice 2.0, this results in 3 identical drop-down colorby menu items: 'Age Range', 'Age Range', 'Age Range'.) Where? Augur 2.0, presumably what V1 JSON field it picks to use for the colorby drop-down (thetitle
field). Missing feature of bug? Bug. Since we have a V1 field calledmenuItem
we should use this for the menu item (title
field) in Auspice 2.0... (both forauspice convert
and when displaying V1) Solution? UsemenuItem
instead oflegendTitle
when interpreting V1 JSONsIssue: Can't exclude anything from 'colorBy' that's in your node-data files (ex: I ran
augur traits
on 'host' but now I don't want it to show as a colorBy anymore, but I do want to include 'country' and 'region' which are in the same 'traits' JSON.) Where?augur export
for V2 JSONS. V1 only includes traits specified inconfig.json
. V2 adds all traits in node-data files, plus any extra given by--extra-traits
argument. Missing feature or bug? Missing feature. We don't have any way to say not to display a trait in a node-data JSON as a colorBy. Solution? Need a way to specify this - not obvious best idea? (This is related to first Clade issue below, but below is a special case we should handle specifically.)(Related - everything is also a filterby. Less pressing, but maybe one doesn't want this for all traits. Also a problem in
augur export
V2 JSONS)Clades
Issue:
clade_annotation
appears in colorBy. This is produced byaugur clades
to allow text-labels on branches, and should never be a colorBy. (clade_membership
should be the colorBy) Where?augur export
for V2 JSONS. Same as above Traits issue - auto-adds all traits in node-data JSONS passed in. But in this case, this should NEVER be a colorby (it only colours one internal branch per clade) Missing feature or bug? Bug. We should auto-excludeclade_annotation
from being a colorby inaugur export
V2. However, it needs to still needs to be on the nodes intraits
! (Orlabels
- see related issue below!) Solution? Auto-excludeclade_annotation
from colorby inaugur export
V2, but still export as node trait/label.Issue:
clade_annotation
isn't displayed as text labels on nodes on the tree. (In fact, there are no Branch Label options available in the left-hand menu.) Where? A little unsure. V1 JSONS have this asattr
clade_annotation
on the nodes intree.json
, and this displays fine in Auspice 2.0. V2 JSONS fromaugur export
have this astraits
clade_annotation
and it does not appear in Auspice 2.0. In V2 JSONS generated from V1 JSONS, (viaauspice convert
) this info is underclade
inlabels
on the nodes, and displays fine in Auspice 2.0. So maybe both, depending on how we want to fix?augur export
V2 should create thislabels
field? Or, Auspice 2.0 should be able to pull this info out of V2 JSONS as it does with V1? Missing feature or bug? Bug. V2 JSONS not getting or pulling out node labels. Solution? Unsure -augur export
V2 should either makelabels
, asauspice convert
does, or Auspice 2.0 should be able to pull info out as it does for V1 JSONSTraits/Geo
Issue: You haven't done
augur traits
on 'country,' so it's not in 'traits' node-data JSON. You want to setgeo
tocountry
(--geography-traits country
inaugur export
V2). Turns out you also must include 'country' in--extra-traits
argument for this to work. Where?augur export
V2. Missing feature or bug? Missing feature. This is minor, but a redundancy we can easily fix! Solution? Anything ingeo
(--geography-traits
) we automatically add to the 'traits' list inaugur export
if it's not there already (due to being in 'traits' JSON' or in--extra-traits
)Legends
Issue: Legends are ordered differently to V1/Auspice 1.0 even if exact same info is supplied in same order.
augur export
V1 (after recent fix) exports colorby traits that have specified colours in the same order as they appear in thecolors.tsv
file.augur export
V2 exports traits with specified colours in alphabetical order. V2 JSONS generated from V1 JSONS (viaauspice convert
) maintain the order specified in the V1-meta file (so the order in thecolors.tsv
file). Augur 2.0 displays in whatever order colorby traits are specified in the meta or unified JSON. Where?augur export
V2 Missing feature or bug? Could be either I guess, but I'd say bug because it's so different fromaugur export
V1 / Auspice 1.0 behaviour. Users can't overrule the alphabetical ordering, which is at best annoying (see Fig 1 - my pretty rainbow colours!!) and at worse nonsensical (see Fig 2 - my age range order is now 1-5, 18-64, 6-17, <1, >=65, and the corresponding colour gradient from young=red old=blue looks random) Solution? Haveaugur export
V2 export colorby traits in same order as specified incolors.tsv
, like inaugur export
V1Fig 1 - i hate to see you lookin so bad
Fig 2 - oh my eyes :(
More!
I noticed something new? Nucleotide colorby doesn't seem to be working properly? If I click on sites in NT view, either entropy or events, the entropy bar zooms correctly and everything, but the tree doesn't recolor... it acts like there is only 1 base at this position (no variance/variants). This doesn't work for V2 JSONS generated by
augur export
V2, V2 JSONS generated from V1 (viaauspice convert
), or V1 JSONS. Must be Auspice 2.0 problem.