Closed gabsie closed 2 years ago
Just an idea: I believe if the cxgene link is added as an accession it will be displayed in project catalogue (but perhaps this requires a HCA metadata version change)
I created a wireframe
The recommendation here is to store links to Cellxgene / SCEA / UCSC in the project metadata schema
project.supplementary_links
External link(s) pointing to code, supplementary data files, or analysis files associated with the project which will not be uploaded. example:
https://github.com/czbiohub/tabula-muris; http://celltag.org/
The suggested change is to highlight supported URLs in the project catalogue.
@gabs / @ESapenaVentura Please can I have examples of the following urls, so that I can determine how we detect them:
Hi, @MightyAx
So for UCSC browser links: https://cells.ucsc.edu/?ds=fetal-thymus https://cells.ucsc.edu/?ds=lifespan-nasal-atlas CellxGene https://cellxgene.cziscience.com/collections/0434a9d4-85fd-4554-b8e3-cf6c582bb2fa https://cellxgene.cziscience.com/collections/6f6d381a-7701-4781-935c-db10d30de293
SCEA: https://www.ebi.ac.uk/gxa/sc/experiments/E-CURD-98/results/tsne https://www.ebi.ac.uk/gxa/sc/experiments/E-MTAB-11011/results/tsne
SCEA: https://www.ebi.ac.uk/gxa/sc/experiments/E-CURD-98/results/tsne https://www.ebi.ac.uk/gxa/sc/experiments/E-MTAB-11011/results/tsne
Just adding to this, that for the regex, don't always expect the "tsne" part at the end. It won't always end like that!
My current plan is to consider any supplementary link that contains:
cellxgene.cziscience.com/collections
as a cellxgene link
ebi.ac.uk/gxa/sc/experiments
as a Single Cell Expression Atlas link
cells.ucsc.edu
as a UCSC Cell Browser link
We can do more regex manipulation to figure out:
But then we should probably be validating the data on input and storing it separately also. If that's in scope for this sprint it's a much bigger task.
Also, any opinions on the use of Logos versus the following text? cellxgene / Single Cell Expression Atlas / UCSC Cell Browser If you would like Logo's please provide them.
@gabsie and @ESapenaVentura to take a look and give feedback
Current progress showing filtering and output for SCEA. The code also works for cellxgene and UCSC but no prod data features those in supplementary links.
It seems like we Will need some regex after all, to display the accessions.
My big question then is how to display cellxgene since they are massive UUIDs and this cell of the table is already massively overflowing, perhaps that's where to bring in the logo.
To help with the overflow I'll just use "SCEA" rather than "Single Cell Expression Atlas"
Implemented accession regex
The accession cell is getting very full for some records.
added external link signifier and started using this formatting everywhere (just ENA so far)
Alexie - Feature complete but some refactoring remaining.
The list-of-links component has been complete, awaiting re-review
Deployment in the EBI Web Dev Environment Failed, but I don't know how to follow this up!
The test step that is failing (because yarn is not installed) was added by @jacobwindsor's recent cleanup, removing it for now.
Deployment Successful, ready for user testing: https://wwwdev.ebi.ac.uk/humancellatlas/project-catalogue/
Known Issue: TSV generation shows [Object object] instead of accession. Will fix tomorrow.
Also, any opinions on the use of Logos versus the following text? cellxgene / Single Cell Expression Atlas / UCSC Cell Browser If you would like Logo's please provide them.
I think it's fine with the names for now. Adding the logos would require a lot more work (we would need to contact each of the institutions for their logos etc)
My big question then is how to display cellxgene since they are massive UUIDs and this cell of the table is already massively overflowing, perhaps that's where to bring in the logo.
I think for now we just need to stick with the UUID, cellxgene have no open APIs where we could query a short identifier for the project
I tried to download the TSV for the project that I added the SCEA/cellxgene/UCSC cell browser accessions to test and the resulting TSV had no headers/entries for those accessions, I assume this is the error you comment above or is it different?
Other than that, I've been testing around and adding a couple of accessions to projects and it works great!
@gabsie to check as well.
Project Catalogue in wwwdev not being deployed to production. Operations ticket to track with EBI Gitlab Team: https://github.com/ebi-ait/hca-ebi-wrangler-central/issues/788
Hey Alexie, this works great when I tried the different links. Small edit you might have to do with regards to your formula for getting the cell browser links. I have seen in the sheet that their links can sometimes be:
https://gut-cell-atlas.cells.ucsc.edu https://covid19-influenza-response.cells.ucsc.edu
so maybe you also have to consider links that end in ... cells.ucsc.edu, not only start with that (I can see this displaying fine: https://cells.ucsc.edu/?ds=adultPancreas)
I have now entered details for 4 projects, you can see them by filtering by cellxgene portal
Gabs
Change in review
Ready for review on dev: https://wwwdev.ebi.ac.uk/humancellatlas/project-catalogue/ can release to prod when you approve.
Second component to allow for a bit more links in staging yet - But ready to be released
@gabsie to take a look
This has been checked and is OK!
As a Data contributor/wrangler, I would like to add the link to my project in a supported Analysis Portal (e.g. Cellxgene/SCEA/UCSC browser), so that other scientists can view my data in the visualisation tool
Acceptance Criteria
To move to a separate ticket (and sprint)