Closed alex-d13 closed 2 years ago
That's close to what I had in mind. I think we only need to emphasize that with SimBu we have a gold standard, maybe like this:
Taking a final round of feedback on this one :)
Decided to add the n_cells parameter to panel b and add small headers to the panels
Very nice, Alex! You could maybe try to have a bit lighter colors for the cells and limit font types and sizes wherever possible.
Maybe @federicomarini has some ideas here?
Color-wise: I like the pastel palette above, or even better something with the spectral palette - there's this nice tool online for grabbing hex codes if you want: https://loading.io/color/feature/Spectral-10/ (https://loading.io/color/feature/ if you want less levels, and equally spaced).
"Pseudo-bulk" vs "pseudo bulk", should be homogeneous (also check in the manuscript how you use it?) - I actually think I have seen it most time as "pseudobulk"; the OSCA book (series) has it as "pseudo-bulk".
In section c, the x axis is a little too small maybe?
A note on the scaling factors: Is it meaningful to have them as integer numbers? I am asking because I am used to see them as ~1, at least in the context of size factors e.g. when normalizing with DESeq2
The "Deconvolution" label is a little bit too much on the left? Say, one whitespace away from the middle?
In the label of (a), there's a missing a
in simula
te
Updated colors and typos :)
I think having the scaling values as integers helps to show the point that we try to estimate the total mRNA levels in the cells; when using the number of expressed genes as scaling factor, this is also an integer for example.
I really like the latest one, good work.
Updated version of the overview figure. I added panel a as a figure to show how and why SimBu is used.
Let me know what can be changed :)