Meaning of PosteriorPs()

[malisas]

Hi, I am wondering whether PosteriorPs() returns
(1) the posterior probabilities of response in the stim condition or
(2) the posterior mean of proportions for the stim condition.

The description of PosteriorPs() says it returns
The posterior probability that the samples subjected to the 'treatment', or 'stimulated', condition responded.

However, in the plot.COMPASSResult() description, the following example is given, which makes it seem like PosteriorPs() returns proportions:

## visualize the proportion of cells belonging to a category
plot(CR, measure=PosteriorPs(CR))

Which description is correct (or are they both correct and I am simply interpreting it incorrectly)?

---

Also, Supplementary Figure 11 of the COMPASS paper plots a “Heatmap of background corrected posterior mean of proportions (equation 11)”, which is defined as:

Background corrected posterior mean of proportions (Pi):
max(0, ¯psi − ¯pui)

I am curious as to which which function call would be used to obtain the quantity plotted in that figure. My guess is that it is a transformed call to PosteriorDiff().

Thank you for your help!

[gfinak]

Thanks, Malisa
Indeed, the description is wrong, it returns the posterior proportions of cells in the stimulated samples. The posterior probability of stimulation is the mean_gamma matrix.

[gfinak]

Still need to regenerate the doc.

[malisas]

Thanks for clarifying, Greg!