Rarefaction and alfa-diversity

[tczinha]

Dear all,

I am trying to understand when rarefaction is appropriate. My samples were sequenced to the same target depth, but the number of reads retained after processing—and the number of ASVs—can still differ among samples. Some differences in ASV richness may reflect the sampling environment. Should I rarefy the data before comparing alpha diversity? I am concerned that discarding reads could change the results.

I have also noticed differences in reported Shannon diversity values for what appears to be the same data. When I calculate the index in R with vegan, I get different values from those reported elsewhere. Could this be due to the choice of logarithm base? My understanding is that vegan uses the natural logarithm by default. How should the base be reported so that results can be compared?

I would appreciate any guidance on both points. Thank you in advance!

[xia.lab]

Indeed the rarefaction will (and should) affect diversity in general. It is the decision you need to make. You can compare result use orginal or after rarefaction (see below).

By default, the alpha diversty is computed based on filtered data which including rarefaction (if performed).

MicrobiomeAnalyst is a platform or pipeline including necessary data processing for more robust results. It is not directly comparable to a single R package dedicated for one task. Please refer to the R command history to see what steps are performed before computing the alpha diversity. In this case, choose Original Data will be more comparable to result from vegan