Different result from ancombc 2.0 and ancombc 2.6

[BangzhuoTongUU]

Hi Lin,

Thanks for creating this tool and keeping it updated! I have a rather short question if you have time answeing it:

Am I expected to see a huge difference between version 2.0 and version 2.6 of ANCOMBC2?
I used the same data, same model and the same parameters to run ancombc2 of the two different versions above. However, I have 200 significant features in version 2.6 and nearly 0 significant features in version 2.0.

Also my input is a phyloseq object consisting of a abundance table and metadata, no taxonomy rank info was included. I am mentioning this because I saw warning saying

tax_level is not specified ...

I checked the default parameters for feature/sample prevelance filtering but I did not find difference of them between the two versions, which means that it cannot be because of the default filtering setting that leads to this drastic change.
Could you suggest what else might cause this result and which version would you suggest that is safer to use? Thanks and look forward to your reply. 感恩。

Best,

Ben

[Maggie8888]

Thanks for your question. We're processing the new version of ANCOMBC and will update the new version soon.

[fconstancias]

Any update on this?