Simplify output directory structure

[adamjtaylor]

Our current output structure is as follows

outdir
    workflow.runName
        meta.id
            workflow.runName
                story.json
                thumbnail.jpeg
                minerva
                    index.html
                    exhibit.json
                    Group1...
                        tile1.jpeg
                        ...
                    ...

The intention of the two occurrences of run name were

1. To allow for multiple runs within the same outdir (does not seem very Nextflowy - we should just change the outdir if we want.
2. To allow for versions to be specified by the run name once the assets are staged in our public bucket (this works to a point but does not really get used and provides a very confusing output structure

I'm proposing this is simplified by removing both of these nested run names.
Story.json will also be moved under the minerva directory

The new structure would be

outdir
    meta.id
        thumbnail.jpeg
        minerva
            index.html
            exhibit.json
            story.json
            Group1...
                tile1.jpeg
                ...
            ...

[thomasyu888]

We should make sure that different tower executions will write to different buckets so you're not overwriting anything.

That said, I agree with this simplification.

[adamjtaylor]

Is there a best-practice way to do this? Or OK to leave to good user practice to change the outdir for each run?

[rxu17]

Had a few comments/questions.

Should we also have a README that just briefly describes the output directory structure here?

Also, is there a release versioning system that nf-artist runs under so we can version by that as a top-level folder for the outputs since we are removing the nextflow workflow name? It could be a parameter we enter (or it can be attached to a repo tag?).

[adamjtaylor]

@rxu17 I've added the output structure to the README.md in #9

[adamjtaylor]

@rxu17, @thomasyu888 looking at nf-core for best-practice guidance, I don't see anything about adding any automated versioning or id- or run-name based structure into the output directory. I think we are therefore good to continue with the simplified approach in this PR and leave it up to the user to be aware of their output versioning etc.

[thomasyu888]

LGTM!