ancombc - structural zero

Hi everyone,

I am really enthusiastic of this new plugin. My question is about the structural zero option of the R package. Does the "struc_zero" option of the R package correspond to the --p-conserve / --p-no-conserve option in qiime2 plugin?

Hey @Carlo77,

The QIIME 2 plugin does not include structural zero estimation as it is essentially a parallel analysis track and we felt it would make sense as a separate optional method, it also did not interact in an obvious way with the formula or our reference level setup for categorical coding, so we opted to omit it in this first pass.

The conserve flag matches ancombc's conserve which according to the docs, changes the variance estimator:

logical. whether to use a conservative variance estimator for the test statistic. It
is recommended if the sample size is small and/or the number of differentially
abundant taxa is believed to be large. Default is FALSE.


Thnak you for your answer. Just one last curiosity: why it has been decided to included the option "--p-neg-lb / --p-no-neg-lb"? (If structural zero is set to FALSE that optioon is supposed to be usless)

Hi @Carlo77,

We included the --p-neg-lb parameter for the use-case where structural zeros are not set to false - in which case, this is used to classify taxons as structural zeros using their asymptotic lower bound. This is an optional parameter, since it will not always be used or applicable (just as you mentioned above).

Cheers :lizard:

To avoid confusion, I believe that parameter was from our initial implementation before we dropped structural zero support, and it never got removed. @Carlo77, you would be correct that doesn't really do anything in the current plugin.

1 Like

Thank you both guys, It sounded strange to have --p-neg-lb parameter and not structural zero one. Everything clear and thank you again for all the effort to keep QIIME update.

1 Like

This topic was automatically closed 31 days after the last reply. New replies are no longer allowed.