The step which I can't do... qiime diversity core-metrics-phylogenetic

Thank you Liz!
I did the demultiplexing, and followed the Atacama soil microbiome tutorial, then Moving pictures tutorial,
The step which I can't do...

qiime diversity core-metrics-phylogenetic \

--i-phylogeny rooted-tree.qza
--i-table table.qza
--p-sampling-depth 200
--m-metadata-file metadata.tsv
--output-dir core-metrics-results
Plugin error from diversity:

The rarefied table contains no samples or features. Verify your table is valid and that you provided a shallow enough sampling depth.

Is the problem with the sample depth?

I keep getting the same issue if I try to do it again

I would appreciate any help)

The second try I put on the sample depth 1 and it did work but then I got another issue

Thank you

Thank you everyone for such an useful resource like this forum!!!
I solved the issue.

Thank you)


Hi @alla_sky - can you please share your solution here? This forum is crowdsourced, and your input is greatly appreciated and will certainly help folks in the future dealing with similar questions.


Hello thermokarst!
I would be glad if my experience will be useful for someone)

I did some filtering first

qiime quality-filter q-score \

--i-demux demux.qza
--o-filtered-sequences demux-filtered.qza
--o-filter-stats demux-filter-stats.qza

I just followed the Moving Pictures Protocol Option 2: Deblur, and the next steps in this tutorial

Then got there
qiime diversity core-metrics-phylogenetic \

--i-phylogeny rooted-tree.qza
--i-table table.qza
--p-sampling-depth 100
--m-metadata-file metadata.tsv
--output-dir metrics

I put 100 as a sample depth because it's our half size of the amplicon output.
Approximate size of our end amplicon product (PCR, before sequencing) was around 200 bp without primers and adapters, so I thought it could work. And it did. How good the data that I got is a question)

Thank you for help

1 Like


I am glad you got something that it workable for you :slightly_smiling_face: I would like to offer a bit of input on your analysis and also some additional insight for future readers of this thread, especially since it looks as if there might be a bit of a mix-up regarding sampling-depth and sequence-length.

Sequence length is exactly what you have described, the size of your individual amplicon outputs, measured in base-pairs.

The sampling-depth is unrelated and pretty confusing. It is number of records that you select to subset from each sample found in your data when performing rarefaction, which is simply a normalization technique that ensures that each sample ends up with the same number of records. Why would you be willing to throw out any of your data? For the long answer, check out this video on rarefaction. The short answer is that if you have different numbers of records for each sample, there are a lot of analysis techniques that you cannot use for that set of data. For some types of analyses, you need to have the same number of records for each sample. Particularly for many/most diversity analysis techniques.

Maximizing usable data from datasets while providing normalized data is a difficult problem and right now rarefaction is a standard normalization method used to solve this problem.

As a technique, it has its fair share of criticism and controversy and there is clearly interest in a better alternative, but there isn't one yet :disappointed:, as with rarefaction you are either throwing away data from samples that have more records than your sampling depth or you are throwing out entire samples if they have less records than your specified sampling depth. Essentially, when choosing a sampling depth, the goal is to preserve as much of your data as possible while working with two diametrically opposed sets of considerations: trying to keep as many total samples as possible while trying to keep as many records as possible from the samples from which had a lot of records.

All that being said, rarefaction does get the job done and big part of making sure that that is the case is choosing a proper sampling depth. For a more in-depth discussion of how to choose a sampling depth, this workshop video has a lot of good information in it. I would definitely watch it and work through your analysis again, as a sampling-depth of 100 is extremely low and you could end up with a lot more data by finding a higher but workable number for your depth.

Selecting the proper sampling depth is a confusing and difficult process, but hopefully these two videos can clarify the process some and help you end up with great quality results.


Thank you)
Really useful videos and explanations)

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