Add main figure 5

[jenna-tomkinson]

Add main figure 5

In this PR, main figure 5 for the manuscript and conference presentation was generated, which shows the evaluation of the model on the siRNA data from plate 4. I am prepared for any and all nit-picks on the figure to make it as clear and manuscript-ready as possible! 😄

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[gwaybio]

Figure nitpicks

Panel A

• It's pretty tough to see the thin lines. Rather than distinguish lines based on width, I recommend distinguishing based on alpha.
• Would it be possible to add small red stars as points to the x axis (somewhere? 🤷 ) that points to the doses we visualize in panel C?
• I would recommend shifting the language in the legend a bit. For example Construct should probably be Target and NF1 Construct 1 probably should just be NF1.
• The x axis should probably be rearranged just a bit siRNA concentration (log10 nM)
• I think this would more accurately describe the y axis label:%NF1\n(normalized to scramble)? (writing out norm -> normalized)

Panel B

• I'd simplify the facet label (NF1 instead of NF1 construct 1) -> we'll have to update this if we get more data
• I'm a bit confused what we're seeing here. A couple questions which may illuminate:
  • Why are there so many more scramble points than replicate points
  • What dose of the NF1 construct are we visualizing here? Perhaps it is worth adding all replicate comparisons (all doses) to the dark blue box? I can also see a world in which we show correlations for each dose individually in this panel somewhere
  • Do you know what is going on in the white box? There is an interesting grouping of points that seem to be distributed evenly.
• The legend title Comparison probably would be better to be Pairwise\ncomparison

Panel C

• This is a very interesting visualization! Especially interesting to see the differences in across wells...
• I wonder if a geom_violin() + geom_point() would be more effective than just a geom_box() - we would like to see the underlying distributions of individual cells per well
• Adding the red stars to panel A would be particularly informative for how to interpret this result!
• While I think splitting by well is interesting and we should include, it might be better as a supplementary figure. I can imagine visualizing the trend per dose in the main figure, but splitting out by well in the supplement. We can include a platemap figure as well, since it seems like there could be well position biases (maybe the upper half of the plate has higher WT probability?)

[gwaybio]

More nitpicks (final)

Panel A

• Title case in x axis label
• Construct to Target
• Fix weird lines in line graph. I think this has something to do with the groups variable in the geom_line() call.

Panel B

• call geom_line() after geom_boxplot() to move dotted line in front of box
• Consider adding back the no controls

Panel C

• Title case in both axis labels
• Consider dropping the alpha slightly and increasing the width of the jitter. It's a bit tough to distinguish high density from low density regions

[gwaybio]

Good to merge when you're happy with figure

[jenna-tomkinson]

@gwaybio Thanks for the review! I will be merging now!