https://nf-co.re logo
Join Slack
Powered by
Any suggestions for ways to <collaborate> with <@U...
# scdownstream
p
Any suggestions for ways to collaborate with @Kevin RUE ?
šŸ‘ 1
e
Is the work only related to Bioconductor packages used in nf-core? The scdownstream pipeline mainly uses Python packages for almost all of its various modules, with some exceptions for a few Bioconductor packages: 
decontx
, 
SingleR
, 
celldex
and 
scds
(unless I'm missing some).
k
Thanks for reaching out. While the proposal centers on Bioconductor as a use case, part of the 'FAIR' acronym is 'Interoperable' and we very much want to promote interoperability between single-cell tools, so very much in scope!
Context: ā€¢ BioFAIR pathfinder projects had to fit within 12 months and less than Ā£100k. ā€¢ In this case, I'm the only one in the team actually paid 40% of my time for the next 12 months. Everyone else is "no cost" volunteer collaborator ā€¢ While I use python tools in my workflows, I am predominantly a Bioconductor package developer, which is why Bioconductor is at the center of this project Happy to clarify anything else as needed. I'll make a note to check out the scdownstream pipeline at the next opportunity. I'm teaching all this week and away at a conference next week
j
@Kevin RUE Assume you're aware of the EBI work on this from back in the day? https://github.com/ebi-gene-expression-group/container-galaxy-sc-tertiary A lot of that involved writing CLIs for R tooling to provide a base layer usable by both Galaxy and Nextflow (or similar), e.g. https://github.com/ebi-gene-expression-group/r-seurat-scripts It seems to have been dormant since Pablo and I left, but might have some useful nuggets.
A lot of this would go away if the bioconductor packages had CLIs built in.
k
Thanks for the pointer, @Jonathan Manning I must say that I didn't know/remember that repo, as an irregular user of Galaxy over the years, mostly for the occasional teaching event. It's my first experience as a tool developer, mostly bringing my expertise as a Bioconductor package developer and frequent user of single-cell related packages across R and Python. I'll be happy to revisit this repo to learn from it and compare with current practices. Since you mention Seurat (R but not Bioconductor), I've been using the very nice implementation of it in tools-iuc as a template for my first handful of Bioconductor-related tools.
A lot of this would go away if the bioconductor packages had CLIs built in.
I can feed that back to the Bioconductor community (the Technical Advisory Board will have a perspective on it). I don't want to pre-empt anything here and now with my first naive thoughts on the subject; suffice to say that it doesn't sound trivial.
j
Yeah, there were parallel efforts for a while. At the time we took the view that baking R logic into wrappers (as in your example there) was a problem because then you have to re-write the wrappers in every workflow language, leading to differences in results.
That's why we wrote a few *-cli packages ourselves, though it was a bit of a pain because every change to the CLI required more PRs to get it in the workflows. You will see much of my handiwork across nf-core in R-based modules where I'm doing exactly the thing I don't like- baking scripting logic into workflow components, because I don't want to maintain multiple external scripting layers for the different packages. But it would be SOOO much easier as a workflow developer if I could pass files to a command line.
this 1
FWIW the same thing applies to Python, which is why at the time we also wrote a Scanpy CLI. Anyway, I'm not longer very tied to the single-cell world, so I'll butt out, but I spend so long thinking about this at the time had to comment šŸ™‚
Forgot to say- we even got part way along the lines automatic CLI generation: https://github.com/ebi-gene-expression-group/cli-generator But then we all took jobs elsewhere so it didn't really progress
k
Anyway, I'm not longer very tied to the single-cell world, so I'll butt out, but I spend so long thinking about this at the time had to comment
Appreciated - I come from the other angle of being actively involved in single-cell world and newer to the art of wrapping. So, helpful comments to give me some historical perspective
I'll have to circle back later today or this week. I'm about to jump into a morning session of teaching. Thanks!
šŸ‘ 1
2 Views
Open in Slack
PreviousNext
Powered by