Compare overlap in Venn to shared chao

Hello again with another question:

I am analyzing my lung data and have an interesting question (not an issue or bug).

I have 58 patients, each with two samples collected at different locations in their lungs – call these loc1 and loc2. After the usual preliminaries I build a Venn diagram for each:

mothur > venn(groups=patient001loc1-patient002loc2)

And do this for each subject. I now have 58 Venn diagrams – each tells me the # of species unique to each location, the overlap # of species, and the richness. Excellent.

Now I do summary.shared to get the shared chao statistic for each of the 58 pairs:

mothur > summary.shared(calc=sharedchao, groups=patient001loc1-patient002loc2, all=T)

I now have an observed overlap value for loc1-loc2, and the shared chao of expected overlap, for each subject.

Here’s the interesting question: the overlap in each subject, loc1 and loc2, is a lot LESS than I a priori expected. Let’s say for patient 001, loc1 has 1000 unique species and loc2 has 1200, and the overlap is 10. Just ten. The sharedchao for that comparison is 9. That tells me that the observed overlap is very likely due to chance alone. Now I have 58 pairs of observed versus expected overlap, and would like to know, for the entire dataset, if the observed overlap is just due to chance alone (looks that way from each pair).

Can I just do a paired t-test of observed and expected (sharedchao statistic) overlap? I don’t think that’s proper but can’t figure out why. How would I compare whether the overlaps are same/different than chance alone for the entire group, as opposed to each one at a time?

Thanks in advance.

Thanks for sharing. good luck