Forum:Consider CodeOcean or other solutions for runnable Biostars answers?

Question

[Jeremy Leipzig]

I'm not sure where "Biostars Next" is headed but I believe there has been a long-standing demand for "runnable" solutions to tougher programming questions. From a technical standpoint, this goal is now achievable.

One of the early attempts at this was Radhouane Aniba's coderscrowd.com, which had some tacit support for Python and R, and in many ways was designed from the ground up to enable Q&A and teaching through debugging examples. That site never quite caught on for whatever reason but was ahead of its time in many ways.

One solution available today (but not as Q&A-friendly) is the CodeOcean platform. CodeOcean hosts a number of bioinformatics and genomics "capsules" (data, code and environment) that serve as reproducible runnable analyses to accompany manuscripts. CodeOcean might augment Biostars answers in some important ways.

I posted this solution to https://biostars.org/p/439754 which had spiraled out of control in dialogue

https://codeocean.com/capsule/4796507/tree/v1

The process went fairly smoothly. CodeOcean has good support for Docker and Conda. There is a slight delay between submitting a capsule for approval and ultimately having it made public. The platform is not open or really free, though a published or public capsule doesn't count against the author's monthly free quota in terms of storage or compute.

Would love to hear your thoughts.

[GenoMax]

I can definitely see value in a solution like the one you posted. It does need some work on the part of the poster. And since public capsules don't count towards a posters quota (as you just pointed out in biostar slack) there is no ongoing cost to person posting the capsule. So long as these capsules persist (like GitHub gists seem to) they would be a good addition.

Answer 1

[Jeremy Leipzig]

OK I'm not sure there is much to debate here, but there are couple things we could agree on for standards going forward

Please title your solutions capsule with the post id e.g.

biostars439754

and add the Biostars tag in the metadata field e.g.

metadata_version: 1
name: biostars439754
description: "This is a runnable solution to the Snakemake problem posed in Using wildcards to accept paired-end reads from Snakemake. Note: the fastq files used in this example are simulated and do not represent the OP's files or protected data"
tags:
- Biostars
- Snakemake

Note this is probably not what CodeOcean conceived as their primary use-case, so maybe let's not create snippet capsules willy-nilly. Maybe limit CodeOcean as a solution when it's clear the OP will benefit from its usage.

[Ram]

I think the new bioinformatics.recipe (which will be some sort of super-site to new biostars) will have a platform that can run code.

[Jeremy Leipzig]

ok that's very much in the same spirit