Update snv-callers to be able to be run from places besides the root of repo

[jaclyn-taroni]

⚠️ Review #1419 first! Unfortunately, I can't stack these in earnest because I'm using a fork!

Closes #1406.

The shell scripts in snv-callers made an assumption that you'd run it from the root of the repository. That will be a problem when the script for generating analysis files (run from scripts/!!) is including in an upcoming PR. This PR introduces the following changes to the snv-callers module:

• Set the working directories of the PBTA and TCGA shell scripts to always be where the script is located, and update the paths in those scripts accordingly.
• Remove the assumption that input files will have paths relative to the root of the repository in the Rscripts analyses/snv-callers/scripts/02-merge_callers.R and analyses/snv-callers/scripts/03-calculate_tmb.R. We still need to make use of rprojroot, however, to make sure we're able to source() appropriately.

[jaclyn-taroni]

Okay, two updates since filing:

• 3b40a91 removes short_histology from the TMB file. It's not necessary and we'd rather have folks joining short_histology from the appropriate histologies file, rather than using that one. I'm not 100% sure if all the downstream ramifications are going to get caught until we have v22 CI data with TMB files that do not contain short_histology to be frank!
• 4e53522 adds functionality to use the base histologies file when running for a data release (related to Update logic in modules where analysis files included in the data releases are generated #1419!)

[sjspielman]

When we made the S2 figures, this path got written in during review. To my knowledge figures are the only thing that needs to use these databases after this module. So, we can either revert back the figures path, or save the databases there in the first place.

OpenPBTA-analysis/figures/scripts/supp-snv-callers-panels.R

Line 26 in e72c11d

scratch_dir <- file.path(root_dir, "scratch", "snv-callers-panels")

[jaclyn-taroni]

I split the difference (kind of) in 054616b. It's a good call for modules to have their own directories in scratch/, rather than just hoping their are no file name collisions in scratch/! But I also think snv-callers is more appropriate than snv-callers-panels.

[sjspielman]

Same comment as before for the scratch path.

[sjspielman]

👍