Metastats vs. Venn


I am sorry for my possible ignorance about how metastats actually works but still I have some questions regarding some illogical differences in output between the venn-command and metastats.
My dataset consists of the gut microbiome (454 sequenced) of 6 samples of smoking and 6 samples non-smoking mice.
I have used merge.groups(shared=nameofsubsampleddataset, After that I used venn() to make a venn diagram of the shared and exclusive observed OTU’s. Herewith I find that approximately 60% of the microbiome overlaps between the groups, and both smoking mice as well as non-smoking mice have some 130 unique OTU’s in each treatment.
However, when I apply metastats I only get a 50 some statistically significant (p<0.05) differentially present OTU’s.
I am aware that metastats uses a randomization procedure and t-testing but venn just shows that there are 260 mutually exclusive (!) OTU’s which do not show up in metastats.
The only thing I can figure is that metastats ignores some 0 abundances in certain samples because of abundance in other samples of one of the treatments whilst venn() does not correct here because after merging there are only two groups, however how does merge or venn deals with OTU occurences, does it take a mean, min, max? Can anyone give me some pointers here?

Kind regards,

Frederiek - Maarten Kerckhof
Laboratory of microbial ecology & technology, Ghent University

I suspect that many of those “unique” OTUs are very rare when they do show up. So a difference between an average of 0.0001 and 0.0000 won’t show up. Also, if an OTU is spotty among replicates within a group, that will also make it very difficult to find differences if things are low abundance.