Beta-diversity Rarefaction

[pumpkin]

Hi,

I want to get the beta diversity from the qiime2.
From what I understand, I think the beta diversity should be based on the rarefied table.
Could I just use the rarefied table of alpha diversity?

For alpha diversity, Qiime can draw the alpha rarefaction plot. But for beta diversity, I do not find similar command lines. Do I miss something?

Thanks in advance!

[ebolyen (Evan Bolyen)]

Hi @pumpkin,

First off, sorry for the terribly delayed response. That's my bad, I've had your post in my todo-list for far too long

@pumpkin

Could I just use the rarefied table of alpha diversity?

Absolutely. Did you use core-metrics or did you rarefy and the run alpha methods? In any case, you should be able to take that same table.

@pumpkin

For alpha diversity, Qiime can draw the alpha rarefaction plot. But for beta diversity, I do not find similar command lines. Do I miss something?

There is a command called diversity beta-rarefaction which is analogous to that. It will give you a jacknifed emperor plot which is probably the closest to what you are looking for. But there's also a heatmap showing the stability of a rarefaction depth and a UPGMA/NJ sample-clustering tree you can look at.

Hope that helps!

[pumpkin]

Hi @ebolyen,
Thank for your reply. I do appreciate that!!!

1. YES. I pick a depth based on the alpha diversity rarefactions and then apply it to the core-metrics. It gives me that rarefied table. Besides, it comes with some beta diversity results which are visualizable in qiime2.

2. For the diversity beta-rarefaction, the depth is required. If I apply a depth (even the same depth as alpha diversity), I think the generated tables would not be the same since rarefaction is an even subsampling step. Is that right?

3. Does it means that if we what to do some alpha and beta diversity analysis, it should be based on the rarefied table with a certain depth by the core-metric command line? Do I need to do rarefaction with other depth for the beta diversity?

Thanks!!!

[ebolyen (Evan Bolyen)]

Hi @pumpkin!

@pumpkin

For the diversity beta-rarefaction, the depth is required. If I apply a depth (even the same depth as alpha diversity), I think the generated tables would not be the same since rarefaction is an even subsampling step. Is that right?

That is correct, and more or less the point. What beta-rarefaction will do is use that depth multiple times so that you can understand how rarefaction impacts the stability of your results (since it is a random process). For example the emperor pane of that visualization uses ellipsoids around the typical points to indicate the variability of the point between rarefaction trials.

@pumpkin

Does it means that if we what to do some alpha and beta diversity analysis, it should be based on the rarefied table with a certain depth by the core-metric command line?

If you want to directly compare them, yes, it is probably best to remove that degree-of-freedom by using the same rarefied table.

@pumpkin

Do I need to do rarefaction with other depth for the beta diversity?

I'm not certain what you mean here, are you talking about rarefying the table at a different depth (from the alpha diversity) for beta diversity, or running beta-rarefaction at multiple depths to explore how that changes? You don't need to change the depth, but it also doesn't hurt anything to know how your study results are impacted by choosing different depths. In fact if you have the time/energy that's a pretty good thing to know.

[pumpkin]

Hi @ebolyen,

Thanks for your time!!! More clear to me.