empty barplot at the end

Hello everyone,
This is pretty weird. I have been running Qiime pipeline and everything is running as expected until the last step of getting bar plots with the taxonomy.

  1. I did core matrix phylogenetics using the following command -

qiime diversity core-metrics-phylogenetic
–i-phylogeny rooted_tree.qza
–i-table pet-table.qza
–p-sampling-depth 15000
–m-metadata-file metadata.tsv
–output-dir core_metrics_results

Then after creating and training datasets from both SILVA and Greengenes I tried to create box plot using following command -

qiime taxa barplot
–i-table core_metrics_results/rarefied_table.qza \ (#I used filtered table here#)
–i-taxonomy taxonomy.qza
–m-metadata-file metadata.tsv
–o-visualization taxa_bar_plots.qzv

But my taxa_bar_plots.qzv does not show any bar plots just a empty plot. I tried to save that files .CSV format through Qiime View server but it gave empty CSV file. The bar plot qzv file is not empty though. The size of this file is >9mb

I had to filter my rarefied_table using -

qiime feature-table filter-features
–i-table core_metrics_results/rarefied_table.qza
–m-metadata-file gg_13_99_taxonomy_assigned.qza
–o-filtered-table id-filtered-table.qza

Can somebody please point out what might be happening?
Thanks

1 Like

Try to use a shallower depth in the core metrics command.

How did you rarefy your table? What depth?

Hi, thanks for reply.
I tried using depth at 5000.
It still showing same results

I would try two things:

  1. From the rarefied table, try to visualize the # of features using the qiime command to convert the table from a *.qza to *.qzv
  2. From the table from DADA2/Deblur, I would do a taxa bar plot from that table to see if you get the same results

Can you link the file through drop box for us to take a look at? Ben

2 Likes

Hi Ben,
Thanks for suggestions.

  1. the table from DADA2/Deblur, I would do a taxa bar plot from that table to see if you get the same result

Table from qiime dada2 denoise-paired shows the same results

the rarefied table, try to visualize the # of features using the qiime command to convert the table from a *.qza to *.qzv

This is how the .qzv file looks like for the data

Thanks

@amoltej,
Sounds like a hardware issue. Sounds like your barplot is not rendering, quite possibly because it is too large to display on whatever system or browser you are using. See here:

Try a different browser, a different machine, so somehow collapsing your data to reduce sample counts before viewing.

Good luck!

1 Like

Hi Thanks for the suggestion.
I have been trying to run it on different machine, and platforms. But everywhere I am getting same output - Empty plot

Does anyone have any other suggestion?

what browsers have you tried? can you share the QZV?

I have tried Microsoft Edge, Chrome and firefox. Here is the file

Thanks

Hi @amoltej,
I have taken a look — this is not a hardware issue but rather the barplot is empty because the visualization has no data. This would suggest that, e.g., all features or samples were filtered out of your table.

The provenance does not really reflect the workflow you described above. It looks like you imported a file called “metadata.tsv” as a TaxonomyTSV format file and used that to filter features from the table… it is possible that you filtered out all features, resulting in an empty table. Is “metadata.tsv” a list of feature IDs? Do those feature IDs actually match features in your rarefied table?

In any case, I recommend using qiime feature-table summarize to examine the feature table you use as input to confirm that the table is not empty. If it is not empty, please share the QZV here so that we can examine.

2 Likes

I agree, sorry I took a look and it looks like there’s no data.

I looked at where the data came from in the provenance and it’s a bit weird:

it looks like that the taxonomy bar plots are pulled from several tables? Ben

Just one table (that has been through a few steps to get to where it is). The step @Nicholas_Bokulich pointed out is highlighted here (see the details on the right side):

1 Like

Also, I just noticed another thing in this provenance graph - @amoltej, you appear to be using the rarefied feature table for the barplots. Once you sort out the issue of filtering, you should not use the rarefied table, but rather the unrarefied one — no point in throwing away perfectly good data for a plot like this!

1 Like