Revision: Run TP53 classifier and assess scores at tumor fraction threshold

[sjspielman]

To assess the affects of tumor purity on our results, we should re-run the TP53 score calculation and classifier, considering only samples that have tumor_fraction>= median tumor fraction (https://github.com/AlexsLemonade/OpenPBTA-analysis/pull/1622).

We should then assess how well the following results hold up:

• ROC scores
  • https://github.com/AlexsLemonade/OpenPBTA-manuscript/blob/b0fdf6f24fe9c021f22e8b7da348708f68877ee5/content/03.results.md?plain=1#L194-L195
• activated/lost score difference is NS
  • https://github.com/AlexsLemonade/OpenPBTA-manuscript/blob/b0fdf6f24fe9c021f22e8b7da348708f68877ee5/content/03.results.md?plain=1#L197
• activated/lost expression difference is signif
  • https://github.com/AlexsLemonade/OpenPBTA-manuscript/blob/b0fdf6f24fe9c021f22e8b7da348708f68877ee5/content/03.results.md?plain=1#L199
• overall trends in TP53 score distributions
  • https://github.com/AlexsLemonade/OpenPBTA-manuscript/blob/b0fdf6f24fe9c021f22e8b7da348708f68877ee5/content/03.results.md?plain=1#L200
• TP53 scores of LFS samples -
  • https://github.com/AlexsLemonade/OpenPBTA-manuscript/blob/b0fdf6f24fe9c021f22e8b7da348708f68877ee5/content/03.results.md?plain=1#L203
  • Note that we plainly state that 2 of these samples have low tumor purity of 16% and 37%. These samples would be excluded from any re-analysis, but this caveat sentence is the MS is fine/good to keep assuming this section is also fine to keep.
• TP53 scores for hypermutants samples (if those are included after thresholding, which is something to check!)
  • https://github.com/AlexsLemonade/OpenPBTA-manuscript/blob/b0fdf6f24fe9c021f22e8b7da348708f68877ee5/content/03.results.md?plain=1#L227

Note further that we emphasize some of the TP53 results in the overall MS summary, so we want to re-evaluate these claims: https://github.com/AlexsLemonade/OpenPBTA-manuscript/blob/b0fdf6f24fe9c021f22e8b7da348708f68877ee5/content/01.summary.md?plain=1#L6