[Work in progress!] Docker configuration for genomics workshop

[standage]

During last week's bug BBQ, several related threads all ran together for me.

• upgrading from bwa aln to bwa mem in the variant calling section of wrangling-genomics (Using bwa mem instead of bwa aln in 02-variant_calling.md wrangling-genomics#111)
• selecting a more recent a data set New workshop data set #42)
• make raw data read-only, hidden; symlink or copy data to a working directory; hide irrelevant details from learners (Make raw files read-only, implement a setup command #43)
• using a single software configuration environment regardless of whether it's deployed on the cloud, on HPC, or on local machines (first discussed in Use containers for configuring software environment? shell-genomics#49, then migrated to Use containers for configuring software environment? #28)

On Friday (and a bit on Saturday) I prototyped a Docker image with the software required to run most of the genomics modules. Note: I haven't installed R or Rstudio yet since I'm not familiar with those materials, but this shouldn't be too difficult to add.
This is a work in progress that ties all of these related issues together.

Before I got much further, I wanted to open this up for discussion and feedback.

• Any thoughts on the data set?
• Should we try to handle these issues piece by piece, or does it make sense to address them all at once?

[standage]

cc @tracykteal @ErinBecker

[standage]

Note: the bcftools flags used in the existing materials are outdated. I used this page to inform my updates to the variant calling workflow.

[raynamharris]

Thanks, @standage for the details. I like that you have chosen a new dataset is still from the Lenski experiments. I think this will minimize the number of changes that need to be made to update the story about the data.

As for updating piece by piece or all at once, how interdependent or overlapping are the suggested changes? Like, how many changes have downstream effects that you may as well address all at once?

[standage]

The bwa mem issue and the new data set are tightly coupled. The others could potentially be handled in a more granular fashion if there was good reason to. But they're related enough that I could handle them all together in a single sweeping change. There would need to be a lot of coordination and communication though, as this will affect a lot of different pieces.

[twrightsman]

Given that we now have two methods (AMI or Conda) to set up the software environment (see #28) and because we need to close all PRs before the workbench transition, I will close this; feel free to re-open after the transition if someone feels strongly that Docker should also be an option!