Interpretation on alpha-rarefaction


I have generated alpha-rarefaction curves for my 8 samples and wondering why the plots for group 4 and 6 end at sequencing depth at around 4-5000 while others reach to 7000.

The value for [--p-max-depth] I set was 4000 which is close to the lowest feature count number (~4200) among the samples in table.qzv file.


Hello Zentoo,

Welcome to the forums!

The alpha-rarefaction plugin randomly subsamples your groups at different sequencing depths, then shows you the alpha diversity value at each. When a group contains only ~6000 reads in total, you can still subsample it down to 1k, 2k, 3k, 4k, and 5k reads, but you can't "supersample it up" to 7000 reads. The line ends at the maximum sampling depth for a group.

Good news, though! The lines for group 6 seems to plateau before 4k reads per sample, so while this group had fewer reads compared to some other groups, it will still enough to capture a consistent alpha diversity value for this group.

Thanks for quick answer:)

Then what decide the range of X axis? (in this case 7000)
And can I assume that the above figure has no problem at all (even for publication)?
Thanks again for giving wonderful advises!

Hi @Zentoo - please double-check your command, if you had specified 4000 for the max depth, the command would've only run the rarefaction trials up to 4000, but the plot you shared here looks like it was run up to 7000. You can confirm the value you used by looking at your viz on, clicking on the provenance tab, and reviewing the parameters specified.


This topic was automatically closed 31 days after the last reply. New replies are no longer allowed.