I’m trying out the membership calculators (jclass) in the dist.shared command. As I thought membership is about the presence and absence of an OTU, it should ignore the abundance count in the shared file, right?
So when I tried a dummy data of 7 OTUs, random sequence counts but making sure all samples have at least 1, I still arrived at significant difference in AMOVA. I used subsample =T in the dist.shared command, could this be the reason?
Yes membership is about presence or absence, and does not take abundance into consideration.
If you had some OTUs that only had 1 or a few sequences, and you subsampled then it is possible they no longer retain any sequences and this would therefore affect your membership testing.
What version of mothur are you using, and what OS? There were a few bugs related to the subsampling that were fixed in the most recent releases.
Yeah, that makes sense. Thanks!
I’m using v1.38.0 on windows.
It would be worth updating to 1.39.5 and checking if you get results more consistant wiht what you are expecting.