Dev

[cademirch]

Changes

• Snakefile now runs the whole fastq to vcf workflow #13
  • Note: when executing the workflow in dryun (-n) mode, not all rules will be listed in the summary. This is because create_intervals is a checkpoint, and after that is run, then the DAG is reevaluated to have the correct order of jobs.
    • Added .test/ which contains some test data and config files for the test data. #12
    • Added .github/workflow which contains the spec for the testing Github Action. Currently set to run on each push and pull request to dev.
    • Merged and tested @sjswuitchik commit to replace GatherVcfs with SortVcf.

      Things I've noticed to consider working on next:

    • The workflow has a high barrier to run on user provided data (aka not downloading from SRA/Refseq) as they have to organize their reads and fasta to match how it would have been downloaded. I think it should be a priority to make this more straightforward. Currently testing an idea I've had, suggestions would be great.
    • Some SRAs aren't actually paired, even if the metadata says it is. If the workflow encounters this, it breaks. I've got a fix for this, but haven't included it in dev as it breaks the fastp stats and I have not figured out how to include that too.
    • Having output directories be configurable makes output and input definitions cluttered and relatively hard to read. I think it could be worthwhile to have output directories be hardcoded, except the upper most results directory.

I think that covers most things. Would love any feedback.

• Cade

[tsackton]

Cade,

This is really great. Agree that making things more functional to run on user-provided data is a high priority. I think it would make sense to open a separate issue for this so we can flesh out options and decide what makes the most sense.

Let's discuss hardcoding output on our next call.

Tim