EXTEND: re-analyze trends at tumor purity threshold

[sjspielman]

Closes #1625 This PR re-assess EXTEND scores if generated from only tumors that pass a tumor purity threshold. A couple steps are taken:

• The EXTEND script itself now has an additional flag to calculate scores only from threshold-passing samples. The EXTEND algorithm does appear to use all genes for baseline calculations, so a full re-calculation was necessary here (as opposed to just using existing EXTEND scores but filtering down to only those samples that pass the threshold). Those scores are outputted in the results/ directory for consumption by the new notebook....
• The module run script will now also run a notebook that compares new results to those we report in the manuscript
• There are a few other tiny changes:
  • Removed an extra back tic from tumor purity README that made it through! woops!
  • At some juncture one of the figure scripts got opened in VS Code, and it ate all the EOL spaces. No other changes are in that file.

Everything is broadly consistent with our existing results, but I do find some areas we could rephrase during revisions anyways. Here is the notebook: 07_EXTEND-at-threshold.nb.html.zip

[sjspielman]

@jaclyn-taroni The tumor filtering function has now been incorporated here. I re-ran the module and there were no diffs in any results, so seems good as long as CI agrees 😄

[jaclyn-taroni]

@sjspielman reviewing now - I think something funky is happening with the HTML output that's committed/pushed at the moment. Specifically in the re-analysis 2 section and beyond.

[sjspielman]

@jaclyn-taroni I've fleshed out the docs with more information for these additions, and I added in some details about running the module in general. I'm glad I did this because some docs were missed in #1653 and that script (06) didn't get added to CI 😬 . Should be set now!! That said, I'll note more generally that this module's README could be expanded quite a bit in other ways that are beyond the scope of this PR, and in the interest of staying on task I avoided too many recreational changes there.

I also went back and forth a bit thinking about this comment https://github.com/AlexsLemonade/OpenPBTA-analysis/pull/1657#discussion_r1098928446, and I think that it really does make sense to always make sure that analysis is run. I moved the thresholding script/notebook out of the logic entirely so it'll always get called.

[jaclyn-taroni]

I don't believe you've resolved the issue with the HTML noted in https://github.com/AlexsLemonade/OpenPBTA-analysis/pull/1657#issuecomment-1421054716 at this point, @sjspielman.

[sjspielman]

I don't believe you've resolved the issue with the HTML noted in https://github.com/AlexsLemonade/OpenPBTA-analysis/pull/1657#issuecomment-1421054716 at this point, @sjspielman.

Woops, somehow missed your comment up there about the HTML when I went ahead to re-request review! Fixing it now. Sorry about that confusion! At some point I was testing locally with CI data, and it borked the notebook.

I am also a little confused about why the Fig 4 panels are being update here, which was probably something included earlier that I did not notice!

Hm, great question... Somehow they got opened along the way and VS Code ate all the EOL spaces are the diffs there, and then further somehow I missed it in git status. As to how PDFs got regenerated, well, who knows that's quite the mystery to me. A series of unfortunate events. I'll go ahead and revert all of those.......

[sjspielman]

Ok @jaclyn-taroni, feeling confident 🤞 that this is ready. Plots are reverted and here's the properly-rendered HTML: 07_EXTEND-at-threshold.nb.html.zip