3D PCoA with Emperor - for dummies

Hello everyone!
I recently started using Qiime2 and I love it!

However I was not able to figure out how to get the cool emperor visualization for my data that can be seen
In the Provenance tab of this example, it says you only need one step from the deduplication output (table & sequences) to the 3D pcoa.

However it seems much more complicated to me. When I checked the necessary input for:
qiime emperor plot --i-pcoa [file].qza […other input and otutput options]
it says it needs an already calculated “PCoAResults”-artifact.

So I looked up how to make that and I found:
qiime diversity pcoa --i-distance-matrix [some_file].qza --o-pcoa [out_file].qza
So to get the pcoa file I need a distance matrix. I looked up how to get that and found:

qiime metadata distance-matrix --m-metadata-column [another_file].qza [other options…]
So, now I am stuck here.

On the one hand this long trail of commands seems wrong, because in the example it was just one step from the table & sequence files. On the other hand it might be the only way to generate the pcoa data. In this case I do not understand what the –m-metadata-column file should be. Do I have to generate the numbers for my samples outside of qiime2?

I could not find other entries dealing with this and because emperor seems to be added quite recently I decided to post in this group! Thank you for your help and creating qiime2!

Hi @giant_virus,

I think the command you’re looking for is

qiime diversity core-metrics-phylogenetic

The command assumes you’ve already taken the following steps:

  1. Build your mappling file --> mapping file
  2. Demultiplexed your sequence files
  3. Denoised or clustered your data (OTU picking) --> feature table
  4. Built a phylogenetic tree --> tree

The command then takes your feature table, tree, and mapping file. You will also need to select an appropriate rarefaction depth to run this command. The moving pictures tutorial is a nice description and discussion of the command.

So, QIIME 2 assumes that you’re not doing your microbiome analysis is a vacuum. You should have a metadata file that describes the samples in your study. For instance, if they’re human samples, where the samples came from, who they came from, whether the person had a disease, etc.

Hope this helps.



Thank you for your kind explanation and thank you for linking the moving pictures tutorial! It is well explained in the current version of the tutorial.

Now I have the same problem as described here.

For some reason the feature_ids were changed when I created the tree. The metadata.tsv information gets added to the feature_id’s names. However I am fairly sure that I can solve this myself, so there is no need for you to look into it. I will post the solution later!

Thank you and have a nice day!

An off-topic reply has been split into a new topic: Mismatched feature IDs

Please keep replies on-topic in the future.

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