Revise S7 to include information about RNA-seq batch effects

[jaclyn-taroni]

Not fully captured by the issue title, but here are the three panels we want to add to what is currently Figure S7:

• [x] (A) a bar plot showing distributions of disease type (cancer_group_display), broken up by library prep strategy (RNA_library) – the objective is to illustrate that disease type is unbalanced between poly-A and stranded datasets, so whatever bar plot strategy we think best communicates this works.
• [x] (B) a UMAP plotting samples from both selection strategies on the same coordinates: https://alexslemonade.github.io/OpenPBTA-analysis/analyses/selection-strategy-comparison/01-selection-strategies.nb.html#raw_clustering
• [x] (C) the UMAP plot highlighting sequencing center: https://alexslemonade.github.io/OpenPBTA-analysis/analyses/transcriptomic-dimension-reduction/04-explore-sequencing-center-effects.nb.html#cancer_group_umap

Current panels A-F become (D-I): https://github.com/AlexsLemonade/OpenPBTA-analysis/blob/627ec427ad0a8d9d913e614c9db50546c56d8283/figures/generate-figures.sh#L188-L210

Then this needs to be recompiled.

We may be better served by splitting this up, but best to start somewhere!

[sjspielman]

Noting Panel A is now being tracked in #1694.

[sjspielman]

S7 has been completed and compiled (#1728). This issue is now all set!