Odd PCoA results based bray-curtis distance

[mol (donnie)]

Dear Friends,

I ran qiime2 analysis based on Deblur method. I took the bray curtis and unweighted UniFrac distance analysis, and found that the PCoA plot based bray-curtis distance had an odd sample distribution, that all the samples were in the lines. However, the unweighted unifrac distance method had a normal PCoA results.

[jbisanz (Jordan Bisanz)]

You appear to have the same issue as this post:

Odd PCoA from distance matrix

Greetings everyone, I followed the movingpictures example and had these weird (i.e. too symmetrical) point distribution on my data shown. These are from core-metrics-results/jaccard_emperor.qzv and core-metrics-results/bray_curtis_emperor.qzv. I once saw similar thing like this in R and if I remember it correctly, it was due to the data transformation I did. Therefore, was this also because of --p-sampling-depth parameter? Thank you. Jaccard [jaccard-batch] Bray-curtis [bray-batch]

I.E. for one reason or another there are very few shared ASVs between your samples. It is most likely a technical artifact related to some sort of processing or batch effect but could be biologically possible under some circumstances if you would have good reason to think there would literally be no shared organisms between communities.

[mol (donnie)]

Thanks! I found the reason that different primer lengths among these samples influenced the performance of Deblur.