Closed abought closed 5 years ago
I've added back the field extraction mechanism. It does impose a namespacing restriction- this is still a change to PheWeb core behavior, but it makes it easier to use the same data source with all the other layers that already work that way.
Per the tooltip issue, I've traced this to pointer-events:none
, the tooltip workaround. We're also seeing it with the exac link on genes tooltips in a modern pheweb. Will delegate that one to Peter as this was his workaround and he has more context.
http://pheweb.sph.umich.edu:5003/region/110.12/2:113925484-114325484
Also fixed the phewas scatter plot tooltips. @pjvandehaar , when would be a good time to meet and parcel out the remaining tasks needed to move this forward?
Great to see this happen. It would be great for PheWeb to serve as showcase of what locuszoom can do!
Sent from my iPhone
On Sep 25, 2018, at 3:24 AM, Andy Boughton notifications@github.com wrote:
Ticket: #17
Purpose
Add a GWAS catalog plot to PheWeb, based on an upcoming LZ.js release
TODO
Recommendations for discussion
Add test data that has a sample region with SNPs in the catalog The catalog track only labels SNPs that are in high LD. Consider adding the "Make LD reference" link back to Pheweb assoc scatter plot tooltips, to encourage exploration. Known issues
The PheWAS coloring mechanism has changed; update this so phewas plots are colored correctly GWAS catalog rug track tooltips are not properly clickable (can't click the GWAS catalog link; investigate pheweb overrides) Since this updates LZ.js to the newest version, generally review app for breakages The layout is slightly off kilter; revise and generally tweak positioning (some "spacer" panels were removed in the layout reconciliation, as their purpose was unclear) Summary of changes
Replace in-repo LZ.js with a version from a CDN DRY layouts to use override mechanism Region plots now use "association_catalog" instead of "standard_association" for their base plot layout (and appropriate child layouts, with modifications) Change association source: remove override of a deprecated function, in favor of explicit namespaced field requests (revisit this) How to test this
Version 0.9.0 has not yet been released yet; we'll need to do that before PheWeb can be released! I've been testing with local static files, representing the tip of the LZ.js develop branch (0a0df23) You can view, comment on, or merge this pull request online at:
https://github.com/statgen/pheweb/pull/112
Commit Summary
Use CDN version for LZjs, and remove locally stored version. Updates for LZ.js 0.9 Reconcile duplicate region plot layout with LZjs core overrides File Changes
M pheweb/conf_utils.py (1) M pheweb/serve/server.py (2) M pheweb/serve/static/region.js (610) M pheweb/serve/static/variant.js (29) D pheweb/serve/static/vendor/locuszoom-0.5.6-locuszoom.css (792) D pheweb/serve/static/vendor/locuszoom-0.5.6.app.js (8892) D pheweb/serve/static/vendor/locuszoom-0.5.6.vendor.min.js (6) M pheweb/serve/templates/gene.html (8) M pheweb/serve/templates/region.html (7) M pheweb/serve/templates/variant.html (9) Patch Links:
https://github.com/statgen/pheweb/pull/112.patch https://github.com/statgen/pheweb/pull/112.diff — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or mute the thread.
No word from Peter as yet. In the interests of meeting ASHG deadlines, and recognizing that it's already been ~10 days, I'm going to put go ahead and make most of the layout and behavior changes originally marked "for discussion", plus a few other tweaks.
Some of the things I held off on could introduce possible regressions, but we can discuss the details after the fact. Will update more tomorrow.
Tooltip functionality changed to support more granular hyperlinks (not just "click on dot to go to phewas").
Still need to identify test data with SNPs in the GWAS catalog; Peter would be more efficient at adding this. This concludes my current queue of work for PheWeb integration until we've looked over the direction and design plan.
Thanks Andy for all your work on this feature. Peter, when are you able to address this PR?
I've pushed (most of) the changes from discussion branch https://github.com/statgen/pheweb/compare/8cdb7...pull-112-head#diff-007e08bf75bfe53b85b14034803e369c . These are intended as a companion to statgen/locuszoom#150.
I'll merge the LZ.js PR tonight barring any further requests.
A note: tests are passing locally (after incorporating the test_all
fix- thanks!)
They are failing on travis even after restarting the build, but the error message is opaque. Peter, have you seen that failure log before?
I’ll try a different gencode url later.
Gencode url? I might be missing context here- what sort of different data do you need?
(Would any of the pending LZ data source changes be of use?)
LocusZoom 0.9 was officially released in late December, containing several of the items requested/used by PheWeb.
What is the status of this PR as a possible candidate for merge? Does it need updating?
Given the historical timeline, I'll hold off on making revisions until I hear back, but help is available on that basis.
This is now on master
.
Ticket: #17
Purpose
Add a GWAS catalog plot to PheWeb, based on an upcoming LZ.js release
TODO
Recommendations for discussion
Known issues
pointer-events:none
is still needed, and if removal is a regression.Summary of changes
How to test this
develop
branch