Chao1_CI error non-numerical value in the qza?

Hi there,

I'm trying to figure out how to visualize the Chao1_ci artifact using the following commands (using qiime 2018.11):

qiime diversity alpha \
  --i-table $input_table \
  --p-metric chao1_ci \
  --o-alpha-diversity chao1_ci.qza

qiime diversity alpha-group-significance \
  --i-alpha-diversity chao1_ci.qza \
  --m-metadata-file $metadatapath \
  --o-visualization  chao1_ci.qzv

The visualization command generates the following error:

ValueError: Non-numeric values detected in alpha diversity estimates.

I found another post and bug report on the forum, but it doesn't seem to provide the same error. I think the problem might be that my metadata is categorical and not numerical. Could this be the issue? If not, do you have any advice on how to use this metric?

I have attached the metadata file, input_table.qza and chao1_ci.qza in case you need to reproduce the error.

Thanks for your help :slight_smile:
table.qza (45.1 KB)
metadata.txt (7.1 KB)
chao1_ci.qza (32.1 KB)

Courtney

Hi @courtneystairs,

The problem is that chao1_ci reports confidence intervals, not a single value. So your data look like this:

	chao1_ci
RE-1820-48	(45, 45)
RE-1820-38	(110.00128107900554, 111.61279335119133)
RE-1820-39	(123, 123)
RE-1820-45	(41, 41)
RE-1820-49	(97.002014832271243, 98.938746989395156)
RE-1820-46	(32, 32.0)
RE-1820-40	(161.00358630666423, 163.17088717082984)
RE-1820-41	(116.00090678158192, 117.40663444153361)
RE-1820-55	(96, 96)
RE-1820-43	(155, 155)
RE-1820-54	(97.000862206484769, 98.379090349111252)
RE-1820-50	(66, 66)
RE-1820-44	(151.00448757096697, 153.75108273567827)
RE-1820-51	(57, 57)
RE-1820-42	(164, 164.00001187418775)
RE-1820-53	(101, 101.00000052748854)

The data are numeric but cannot be parsed by the alpha-group-significance method, since they appear as tuples.

Oh! I see, thanks!

OK so it appears that I cannot convert it to a .qzv file. I suppose its best to just export it with the qiime export function into a TSV file (like what you have made)?

Thanks! Very helpful. :slight_smile:

You should be able to use qiime metadata tabulate to view as a QZV. But you will want to export anyway if you want to manipulate these data in any way (e.g., maybe you could convert the CI to the average, then reimport to QIIME 2 for comparison between groups?). Chao1 CI is a bit unique — all other alpha diversity metrics should be compatible with alpha-group-significance.

e.g., maybe you could convert the CI to the average, then reimport to QIIME 2 for comparison between groups?

Thanks Nicholas! Sounds like a good idea :slight_smile:

Chao1 CI is a bit unique — all other alpha diversity metrics should be compatible with alpha-group-significance .

You’re right! No other problems so far.

Thank you!

1 Like