Rarefraction curve is so confusing in QIIME2?



Hi, developer,

Thank you for your time. After reading all posts related to the rarefaction curve posts in the forum. Some questions still confusing for me.

I got a rarefaction curve in QIIME2, which is so confusing. Based on my understanding, the curve converges to the flat curve is the symbol of enough sequencing depth.
But in the plot of QIIME2 rarefaction curve. No matter how few reads the sample has, the curve will reach converge.

In my sample, J1 and F1 are exactly the same DNA, which I sequenced twice, while F3/J3, F5/J5 are in the same fashion. From QIIME2 produced curve I really do not know how to distinguish from them, which is better.

Enclosed is the qzv file I analyzed. May I have some suggestions on how to diagnose the sequencing depth is enough or not from the QIIME2 produced curve?? From my current result, which sample is enough sequenced?

Appreciate your help!!! Tons of thanks!

alpha-rarefaction.qzv (636.5 KB)

(Nicholas Bokulich) #2

This is not a feature specific to QIIME 2 rarefaction curves — this is a feature of your data. It indicates that your samples likely have different absolute values of observed_otus, since the asymptote is (theoretically, but not really) the true value.

So you would use these results to choose a rarefaction depth in the same way that you would any other time… the only problem for you is that some samples appear to converge later than others, so you will need to decide whether you want to lose samples or lose coverage of those high-diversity samples.


Hi, @Nicholas_Bokulich,

Thank you for your feedback. Really thankful!
But it is still really so confusing for us to use the rarefaction curve to check the sequencing depth is enough or not. I still have no idea.

May I please know is there any other method in QIIME2 can be used to check the sequencing depth is enough or not? Since from current rarefaction curve, it is hard for me to diagnose the sequencing depth.

Thank you so much.


(Nicholas Bokulich) #4

Hi @Brandon,
You can use the --p-max-depth parameter to adjust the maximum rarefaction depth shown in that plot. That may make it easier to visualize the samples that have fewer samples. You can also always export the data and plot in another program where you have more control over axis scale, etc.

Good luck!


Thank you @Nicholas_Bokulich