search

AlexsLemonade / OpenPBTA-analysis

The analysis repository for the Open Pediatric Brain Tumor Atlas Project
Other
100 stars 67 forks source link

Figure 5 with larger text #1751

Closed sjspielman closed 1 year ago

sjspielman commented 1 year ago

Closes #1740 Found another zenodo \n problem in this PR, fixed. 😭 - Part of #1744

I'm opening this PR for Figure 5 before it's done to get some feedback, hence Draft. I've updated panels A-C so far, and it feels really busy to me but I'm not sure that's entirely avoidable if we want larger legends! As I wrote in #1740, another approach for GSVA labeling could be to use the cancer group abbreviations to directly label the bar. Is this worth exploring? Feedback very welcome! @jashapiro @jharenza

sjspielman commented 1 year ago

@jashapiro At a minimum I'm pretty sure I've fixed the PNG stripiness. Here's another partial compilation, again only first 3 panels. This time, the UMAP legend is below and the heatmap legend is above their respective plots. I do think it looks better, though still (unavoidably?) busy.

jashapiro commented 1 year ago

@jashapiro At a minimum I'm pretty sure I've fixed the PNG stripiness. Here's another partial compilation, again only first 3 panels. This time, the UMAP legend is below and the heatmap legend is above their respective plots. I do think it looks better, though still (unavoidably?) busy.

I definitely like that better. One more little idea: We could eliminate the UMAP ticks and values... Just keep the UMAP1 & UMAP2 axis titles, but claim back a bit of space there and remove some clutter.

jharenza commented 1 year ago

Panel letters are missing here and looks like a cluster branch is also missing in B. Another possibility to increase pathway text in B could be to lengthen the heatmap and make C into 3 columns and 4 rows. This could also give you some more space to space out the legend in B from the heatmap.

sjspielman commented 1 year ago

This PR is now officially ready for real review. In https://github.com/AlexsLemonade/OpenPBTA-analysis/pull/1751/commits/9f09ad52e9d34d0d241b82e39a4820ec53e78bd4, I have a fully compiled updated Figure 5 for feedback! There is lots more space in between A/B which I think is helpful to make the overall figure less busy. While making this, I also explored swapping A and B (as in, heatmap on left and UMAP on right) to better line up with D and E, but that alignment did not work as well as I'd hoped, so I kept the panel arrangements as is.

sjspielman commented 1 year ago

All code changes here are in figures/ so it doesn't run through CI. Can merge before checks.