-
Notifications
You must be signed in to change notification settings - Fork 31
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Shear UI #505
Shear UI #505
Conversation
merged @ElDeveloper's suggestions Co-authored-by: Yoshiki Vázquez Baeza <[email protected]>
The following artifacts were built for this PR: empire-biplot.qzv, empire.qzv, empress-tree.qzv, just-fm.qzv, plain.qzv |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks so much @kwcantrell, looks GREAT and it is useful. Here's a few high-level observations:
- I noticed some oddities with the barplots when you filter down to very few tips in the tree:
Notice the selection of only Bacteria and only Thermi.
Seems like it's also an issue with the rectangular layout:
- Do you think we should include a warning message that reminds users that hiding elements in the tree can potentially be misleading?
- Do you also think we should also include a warning about the ordination not being affected by any of the visibility changes?
- Can we have an easy way to deselect all from a layer?
- The shearing panel shows up in the UI even when there's no feature metadata, see for example
plain.qzv
.
@@ -585,6 +585,13 @@ p.side-header button:hover, | |||
white-space: nowrap; | |||
} | |||
|
|||
.shear-layer-legend div { |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Any reason why this class is duplicated compared to barplot-layer-legend
I notice there's a div
filter but any reason why these need to be two different classes?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
When an overflow occurs I wanted the title of the filter to always be in view. For example if you select "Level 7" for the shear filter, you can scroll and always see "Level 7". This occurs because .shear-layey-legend
only applies to div
elements within the shear-legends
container. If I used barplot-layer-legend
then the title of the filter would also scroll and if I added div to barplot-layer-legend
then it would break the barplot legend.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
There's probably a better solution but that is what was easiest for me.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
A simple-ish solution might be creating a title element for each shear layer (for example, the Layer N
text in barplot layers), and then not having a title at all of the stuff in the shear layer legend. Since each feature metadata column can correspond to at most one shear layer, the title we assign a shear layer could be Shear Layer: Level 4
or something like that?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@kwcantrell Would it be possible to make this change? (edit - whoops, looks like github lost track of the earlier comments -- see parent at https://github.com/biocore/empress/pull/505/files#r603485148)
I don't think it should require much if any additional work, and should reduce the amount of code complexity -- I think it would let us remove this duplicated CSS code.
This would involve storing the "Level N" text in an element with <h3 style="text-align: center;">
instead of in the legend. The code for this is adaptable from the barplot layer JS here.
We could also store the select / unselect all buttons in a <p>
underneath this h3 tag, with the same style as the feature / sample metadata barplot toggle buttons (see code here). This should make the UI look nicer and would be more consistent with the barplot stuff, I think?
I made these changes in a local copy of the repo in a few minutes of editing, and the UI now looks as follows. It's a small change but I think the consistency will help users.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
empress/support_files/js/empress.js
Outdated
keepNames.push(name); | ||
} | ||
|
||
this._tree.shear(new Set(keepNames)); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
What do you think about adding a method to the TreeController class that takes in a map and shears the tree. For example shearWithMap(shearMap)
. This would update the internal sheared tree and we can retrieve the names as needed. Perhaps even packing the !== null
check in the getAllNames
method.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I'm not entirely sure what you mean here. @ElDeveloper do you mind elaborating a bit more? What exactly is the map in shearWithMap
and what do you mean by packing the !== null
check in getAllNames
?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
What I meant is that some of the shearing functionality in empress.js
can be moved to tree-controller.js
. Namely the part where you figure out which tip identifiers to shear. My suggestion is to take the data in shearMap
(the argument that's currently being passed to Empress.shear
) and pass that to TreeController
for example TreeController.shearToMap
(or shearWithMap
up to you). That way additional checks like verifying that names aren't null, etc can all be done in TreeController
rather than in Empress
.
If this doesn't make sense, let me know and we can hop on a quick call.
*/ | ||
Legend.prototype.addCategoricalKey = function (name, info) { | ||
if (_.isEmpty(info)) { |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Why is the error check being removed?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@@ -0,0 +1,208 @@ | |||
define(["underscore", "util", "TreeController"], function ( |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Great to see this module is using MVC. Even though objects are not exported outside of this closure, I would still suggest that you namespace them, for example: ShearingLayer, ShearModel....
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks @kwcantrell
Thanks @ElDeveloper! Those oddities you pointed out are going to be bugs. When implementing the shear I did not really test out the sample metadata for the barplots as I thought it would behave like feature metadata but I guess that is not the case. I'll need to look the code to figure out what is causing it. |
Thanks for looking into it! |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This is a really neat change, thanks! It looks like there are a decent amount of things that could stand to be improved before we throw this at users; let me know if you have any questions.
empress/support_files/js/empress.js
Outdated
this.getLayoutInfo(); | ||
|
||
// Undraw or redraw barplots as needed (assuming barplots are supported | ||
// in the first place, of course; if no feature or sample metadata at | ||
// all was passed then barplots are not available :() | ||
if (!_.isNull(this._barplotPanel)) { | ||
var supported = this._barplotPanel.updateLayoutAvailability( | ||
this._currentLayout | ||
); | ||
if (!supported && this._barplotsDrawn) { | ||
this.undrawBarplots(); | ||
} else if (supported && this._barplotPanel.enabled) { | ||
this.drawBarplots(); | ||
} | ||
} |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Looks like this is duplicated code from Empress.reLayout()
; could you make this into a separate function that both reLayout()
and shear()
call, to avoid having duplicate code cluttering things up?
empress/support_files/js/empress.js
Outdated
var nodeNames = this._tree.getAllNames(); | ||
// Don't include nodes with the name null (i.e. nodes without a | ||
// specified name in the Newick file) in the auto-complete. | ||
nodeNames = nodeNames.filter((n) => n !== null); | ||
this._events.autocomplete(nodeNames); | ||
|
||
this.getLayoutInfo(); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Looks like this is duplicated code from Empress.initialize()
; as with below, could you make this into a separate function that both initialize()
and shear()
call?
throws(function () { | ||
legend.addCategoricalKey("oops", {}); | ||
}, /Can't create a categorical legend when there are no categories in the info/); | ||
}); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Ditto with @ElDeveloper's comment above in legend.js -- I don't see why this should be removed. Does this PR involve the use of "empty" feature metadata categories somehow?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
It is possible to remove all feature metadata categories in cases where all tips where removed from the tree or the remaining tips do not have metadata associated with them.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Actually I think if nodes dont have metadata they get assigned a "N.A." value. But the reason I removed it was the use case where all tips where sheared from the tree.
empress/support_files/js/empress.js
Outdated
/** | ||
* | ||
*/ | ||
Empress.prototype.getUniqueSampleMetadataInfo = function (cat) { |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I know you mentioned docs were a work in progress, but this function in particular could really use some details -- I am unsure why exactly this has to be used in place of getObsBy
(I get that this removes tips that have been sheared from the tree, but it isn't clear where this change is useful "downstream" later on in colorBySampleCat
).
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I agree documentation is definitely needed!
Reason this is used in place of getObsBy
is that the BIOMTable
object does not access to the tree so it is unable to remove the nodes and I don't want this to be "downstream" in colorBySampleCat
so we can avoid having to implement this in future functions if they require sample metadata.
added some suggestions by @ElDeveloper and @fedarko Co-authored-by: Marcus Fedarko <[email protected]> Co-authored-by: Yoshiki Vázquez Baeza <[email protected]>
@fedarko I am glad you caught the visual bug (4). I had changed the way I also registered a shear observer for the emperor callbacks so that they get updated when a user shears the tree. The observers will also be unregistered after the 4 second timer expires. w.r.t. the animations, its get a bit hairy to properly deal with the shear functionality during an animation so instead I added a warning message to the animation panel. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
A few updates based on the changes made. Sorry to be a pain and thanks so much Kalen!
<p class="side-panel-notes"> | ||
<b>Warning:</b> Do not shear the tree while an animation is running. Please | ||
stop and restart if the tree was sheared during an animation. | ||
</p> |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks for adding this -- looks good!
If we are going to completely say that the user shouldn't do shearing during an animation, I think it might make sense to just explicitly prevent this -- I assume this warning would also apply to synchronized animations that Emperor starts?
Maybe we could have the Animator
class update the Empress
class somehow (so that the Empress
class always has knowledge of if an animation is ongoing), and then ShearModel
could use that knowledge to only apply shearing changes when an animation isn't ongoing. Or, IDK, you would know best how to do that :)
...We don't have to do this before merging, tho -- I recognize that this PR has been waiting for a while, so I'm ok if we want to get this in and then take care of preventing this case later.
@fedarko I went ahead and added a bunch of code to disable sample/feature metadata coloring, barplots, and shear tree while an animation is active. Since I added some new code, I am sure you can continue to be "pain" 😆 (please continue to be a "pain" as that will ultimately lead to a better feature). This is how the UI looks while an animation is active: Additionally, there was another case where the user started the animation using the emperor interface. In that case, I also disable the |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thank you so much, @kwcantrell! I'm really excited by these changes -- this is very comprehensive. I think this PR will actually also close #195, an issue that it turns out we opened exactly one year ago (!!). So that's neat.
All I have at this point are minor suggestions. One more round and I think we can merge this in!!!
It isn't necessary by any means, but I also think we can reduce the size of the warning messages for the disabled tabs a bit -- we can do this by adding another flag variable, is_empire_plot
, that is passed from core.py
to empress-template.html
to the EnableDisableSidePanelTab
objects. This flag variable will control whether or not the note about Emperor animations will be shown in e.g. the sample/feature metadata coloring tabs while they're disabled (so, if there isn't an Empire plot, then we can just show a smaller warning). If you're ok with it, I propose the following changes to three files within the code:
I've tested this on a checked-out version of this branch, and I think this should work. The differences between these three files and the repo can be assessed by copying them in and using git status
(note that my other suggestions in this review should be applied on top of these changed files).
Thanks again, and let me know if I can be of further help.
added suggestions Co-authored-by: Marcus Fedarko <[email protected]>
Reason for removing \t characters is that these were the only files in the Python / JS codebase that used literal tab characters (instead of ordinary spaces).
The other tabs don't use bold for "This tab" and "To re-enable this tab", so we don't use it here either.
Thanks for the many reviews @fedarko! 😆 This feature went through a lot of iterations and I think the final product is way more polished as a result. |
Thanks so much @kwcantrell (and @ElDeveloper)! It's really great to finally have this in :) At long last, I guess we can push out a release early next week? 👀 👀 👀 |
Its definitely time for a release. Although, if we can merge #484 , #521 (updated to include #484) and the new adjust color alpha feature first, then I think we will truly have the "next gen" release! 😲 I believe #484 and #521 just have a few issues left to address and the adjust color alpha is ready for review (I was just waiting for #505 to be merge before I opened the PR as it abstracts/builds on #505). However, I am totally fine with cutting a release next week! |
The idea of Next-Generation EMPress ™️ certainly sounds enticing ;) However, I'd vote we cut a release ASAP, just because the review process can sometimes take a while (... as I guess was demonstrated by me on this PR 😅 ), and it's been some time since our last release. We could always cut one release next week, and then cut another release soon afterwards with #484 / #521 / alpha adjustment / etc. I have a few features in prep that I hope to get ready soon as well (#322, #371, #504). |
Okay, that sounds like a plan then! Lets to cut a release next week! |
I still need to finish documenting and add a couple of tests but in the meantime if @ElDeveloper and @fedarko can check the .qzv for UI bugs (i.e. test different combinations of coloring/shearing) to see if you can find errors that I did not count for that would be great! This PR branches from tree-controller so #492 needs to be merged in before this PR.