Save figures as pdfs

[Bernadetadad]

Currently non-altair figures (such as the ones in build_codon_variants notebook ) are saved as .png indocs/_images. Can we export figures as .pdf (or .svg) instead, this allows them to be subsequently modified in illustrator/inkskape as a vector graphic.

[jbloom]

Making a note, this may be helpful when trying to sort this out: https://nbsphinx.readthedocs.io/en/0.8.9/code-cells.html#Plots

[jbloom]

@Bernadetadad, I can't get this to work in a general way for all plots. What are specific plots of interest here?

[Bernadetadad]

I'm mostly thinking about Fig1 for paper so I think Average mutations per variant and functional score distributions is what I need.

[Bernadetadad]

@jbloom could you have a look at this, I'm currently writing the first section and would be good to have these for making figures.

[jbloom]

@Bernadetadad, does something like attached work for average mutations?

For the functional scores, I sort of think a ridgeline plot might be better than a violin plot as it won't emphasize the outliers so much. Would you agree on that?

[Bernadetadad]

yes on both points, I'll probably still play them up in illustrator but this works

[jbloom]

@Bernadetadad, for the functional scores, what about a ridgeline plot like this?

I have just made on per-library (averaging over all replicates for library), as I thought that might (?) be better way to show for paper. Does this seem about right?

I typically don't like to use irrelevant color scales, but we could also color in the following ways:

Arbitrary colors per category:

Colored according to the mean effect of mutations:

I sort of like the last coloring scheme (mean effect of mutations) best, but what do you think?

[Bernadetadad]

So much better than the violin plots we had! Is it possible to add a legend for colors in the the bottom one?

[jbloom]

I could add a legend like this (cleaning up the label). Do you think that is helpful? I can't quite decide: it's sort of intuitive without and maybe we don't need to know the mean. But perhaps it is nice to show this if people want to sort of be able to see what the mean is for each. I'll add it if you think helpful.

[jbloom]

I'll add a legend, and you can always remove it in Illustrator if you want.

[Bernadetadad]

I kind of like the legend because then you don't need to read the text to know what the colour scheme is calculated on, it might be obvious for us that it's mean but not necessarily for people seeing this the first time.