Qiime2-2018.8 categorical variables skipped in diversity analysis?

User Support
1

hello again...

in version 2018.6 I was able to run diversity analysis with my metadata file. I've attached my metadata file used in 2018.6 and the diversity qzv files for faith-pd-group-significance and evenness-group-significance. In these qzv's when I visualize them the categorical variables "Primer", "FermHourCat" and "Description" is present and I can view the results for the diversity analyses.

In the 2018.8 version - I ran the analyses as I would in the 2018.6 version with the same metadata file and per the moving pictures tutorial and this time my Primer, FermHourCat and Description variables are not being recognized as a categorical variables. On the visualization it states:

" The following categorical metadata columns have been omitted because the number of groups was equal to the number of samples, there was only a single group, or the column consisted only of missing data: Description, FermHourCat, Primer"

Is there something I'm missing such that my metadata file columns are not being recognized the same way in the newer version analyses versus the older version?

Version 2018.6 commands:

(qiime2-2018.6) wsb255bioimac27:Issue_0031_q2-insertion-bulk mel_local$ qiime diversity alpha-group-significance --i-alpha-diversity core-metrics-results/faith_pd_vector.qza --m-metadata-file fixed-metadata.tsv --o-visualization core-metrics-results/faith-pd-group-significance.qzv

(qiime2-2018.6) wsb255bioimac27:Issue_0031_q2-insertion-bulk mel_local$ qiime diversity alpha-group-significance --i-alpha-diversity core-metrics-results/evenness_vector.qza --m-metadata-file fixed-metadata.tsv --o-visualization core-metrics-results/evenness-group-significance.qzv
Saved Visualization to: core-metrics-results/evenness-group-significance.qzv

Older Version - 2018.6 files
fixed-metadata.tsv (17.4 KB)
faith-pd-group-significance.qzv (333.8 KB)
evenness-group-significance.qzv (330.7 KB)
evenness_vector.qza (39.3 KB)
faith_pd_vector.qza (39.7 KB)

Version 2018.8 commands

(qiime2-2018.8) wsb255bioimac27:Issue_0033_cutadapt_qiime2 mel_local$ qiime diversity alpha-group-significance --i-alpha-diversity core-metrics-results/faith_pd_vector.qza --m-metadata-file fixed-metadata.tsv --o-visualization core-metrics-results/faith_pd_group-significance.qzv

Saved Visualization to: core-metrics-results/faith_pd_group-significance.qzv

(qiime2-2018.8) wsb255bioimac27:Issue_0033_cutadapt_qiime2 mel_local$ qiime diversity alpha-group-significance --i-alpha-diversity core-metrics-results/evenness_vector.qza --m-metadata-file fixed-metadata.tsv --o-visualization core-metrics-results/evenness_group-significance.qzv

Saved Visualization to: core-metrics-results/evenness_group-significance.qzv

Version 2018.8 files
evenness_group-significance.qzv (317.4 KB)
faith_pd_group-significance.qzv (317.9 KB)
fixed-metadata.tsv (17.4 KB)
faith_pd_vector.qza (50.2 KB)
evenness_vector.qza (50.0 KB)

Thanks!

2

Hey there @mmelendrez!

You rarefied at two very different depths between the two analyses --- first at 5000, the second at 8000. The second depth only has 45ish samples, while the first has 131ish. Now, imagine what your metadata looks like in terms of the warning you provided above with 90 or so samples removed - make sense?

Keep us posted! :qiime2: :t_rex:

2 Likes
3

Oh interesting - when I looked at the demux file I thought i was actually looking at/keeping most of the samples, I must’ve missed the load of the rest of the qzv on chrome! Good to know - I’ll go back and re-rarefy! Thanks…

1 Like
4

Ahh was looking at the wrong table :woman_facepalming:... - wow - using cutadapt to trim everything to the same size per primer sites really slaughters my resulting sequences after clean up!
allprimers-cutadaptTrim-table.qzv (667.1 KB)

My entire Pro dataset is essentially gone...guess it makes sense I essentially cut off 174 bps off the front of the Pro sequences to make them match up with the primer start sites of Arc and V4 and in looking at the demux - the 'before' trim looks great...
Pro-pe-demux.qzv (287.4 KB)

The after trim...took a dive off a quality cliff in what's left and I assume dada2 just axed most of it...
Pro-pe-demux-primerTrim.qzv (292.6 KB)

:no_good_woman::woman_shrugging:

Good to know. q2-insertion (rather than cutadapt to primer sites) is the way to go (at least with these datasets) when analyzing fragments of different lengths. Knew that going in but now I can show the PI and it adds to the discussion from these forum topics:

2 Likes
5

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