Is a pseudocount necessary for q2_ANCOM-BC?

Please read the following before posting!

Hello, Is it necessary to use a pseudocount when running data in q2-ancombc? I am not getting any error and results appear to be complete without pseudcount. Running with pseudocount gives similar results.

However, when I generate differential plots in q2-QURRO, the sample plot is not being generated. It seems all taxa are represented in the feature plot, but I receive the following error for the sample plot:

0 / 20 samples (0.00%) currently shown.
20 / 20 samples (100.00%) can't be shown due to having an invalid (i.e. containing zero) log-ratio.

I would be very grateful for any suggestions! Thank you, Nate

Hello Nathaniel,

I think ANCOM-BC does this internally.

I guess we should also discuss why pseudocounts are needed. I like the discussion in this document:

Note that the log-ratio transformation, by its nature, fails if the input data contain any zero values. By default, this function replaces all zero values with 1. Alternatively, the user may set the parameter alpha greater than zero to approximate log-ratios in the presence of zeros (via the Box-Cox transformation). However, the topic of zero replacement is controversial. Proceed carefully when analyzing data that contain zero values.
(emphasis mine)



Thank you, Colin! I thought this must be the case and appreciate your help
Cheers, Nate