Update cell-type-wilms-tumor-06 workflow

[sjspielman]

Closes #824

This PR updates the cell-type-wilms-tumor-06 workflow file to:

• Use Docker (note that I did also confirm just to be sure that the system dependencies needed are indeed already in the Docker container, thanks bioc base image!)
• Use a subset of samples only when testing on PRs. I chose these samples as the ones that are specifically run through copyKAT & inferCNV, since the workflow currently evaluates these methods only on a subset of samples. This way, those samples will be present! There is nothing else particularly special about them.

[sjspielman]

Ah, so this is the first workflow we've used Docker in where the container doesn't have a conda environment, so aws bells and whistles for downloading data aren't installed. I think the move is probably to add a step in the workflow to build the openscpca conda environment, vs adding it to Docker. @jashapiro, any other opinions on this?

[jashapiro]

Yes, I would rather leave the docker image alone. I don't think you should need to install conda; just a step to install aws directly should be sufficient:

curl "https://awscli.amazonaws.com/awscli-exe-linux-x86_64.zip" -o "awscliv2.zip"
unzip awscliv2.zip
sudo ./aws/install

[jashapiro]

This looks good with just a couple minor naming/style comments.

A separate thought is if we want to track the fetusref.SeuratData package in renv and not install it as part of a script as we currently do. This would obviously make the docker image bigger, but save repeated downloads. (Other reference files could also potentially be installed as part of the Docker image, but this one is the easiest given current infrastructure.)

[jashapiro]

The changes to code that will affect the docker image should really be done in a separate PR... I'm sorry I was not clear about that!

The reason is that the docker image has to be rebuilt and repushed, so here you are not actually testing with the new docker image (which is why this failed... the current docker image still needs you to install within the script)

[sjspielman]

The changes to code that will affect the docker image should really be done in a separate PR... I'm sorry I was not clear about that!

Yeah, I realized this seconds promptly after pushing but took the dog for a walk first! Going to revert here :)