Error grouping metadata column for relative abundance taxa barplot – issue retrieving 'sample-id'

[anna.abbey]

Hi everyone,

I am very new to QIIME 2 and trying to analyze bacterial community data. I am trying to group my samples and generate a stacked bar chart showing the relative abundances of taxa. However, I am encountering an error when attempting to group the data using the qiime feature-table group command.

Here is the command I used:

qiime feature-table group
--i-table dada2_table_filt_contam_unclass_removed.qza
--m-metadata-file Abbeyv4v5_sample_1metadata.tsv
--m-metadata-column sample-id
--p-axis sample
--p-mode sum
--o-grouped-table dada2_table_filt_contam_unclass_removed_sample-id.qza

And this is the error I receive:

There was a problem with the command:
(1/1) Invalid value for '--m-metadata-file': There was an issue with
retrieving column 'sample-id' from the metadata.

I have confirmed that Abbeyv4v5_sample_1metadata.tsv includes a column labeled sample-id. I have validated this in keemie and used in my dpwnstream analysis untill now. I am wondering if this could be due to a formatting issue I am missing (e.g., hidden characters, header issues, or case sensitivity).

My goal is to group the samples by metadata and then generate a stacked bar chart of taxa relative abundances using the following command:

qiime taxa barplot
--i-table dada2_table_filt_contam_unclass_removed_sample-id.qza
--i-taxonomy 16s_taxonomy.qza
--o-visualization taxa_barplot_sample-id.qzv

Any advice on how to fix the grouping step error or best practices for preparing metadata would be greatly appreciated.

Thank you in advance!
Anna

[colinbrislawn (Colin J Brislawn)]

Hello Anna,

Welcome to the forums!

The qiime taxa barplot command is designed to take your original feature table and metadata file directly to generate the plot. It handles all the necessary grouping and visualization internally.

Try this command using your original, ungrouped table:

qiime taxa barplot \
  --i-table dada2_table_filt_contam_unclass_removed.qza \
  --i-taxonomy 16s_taxonomy.qza \
  --m-metadata-file Abbeyv4v5_sample_1metadata.tsv \
  --o-visualization taxa_barplot.qzv

You would only use feature-table group if you wanted to combine multiple samples into a single new sample (e.g., pooling technical replicates or merging all samples from a treatment group). For viewing individual samples in a bar chart, it's not needed.

Let us know if that works for you!

[timanix (Timur Yergaliyev)]

Hello Anna,

If you are still interested in grouping your samples, in your metadata there should be a new column, that contains information about the group to which each sample belong. It should not be named "sample-id", since such name is reserved for index column. Name it Group, Grouping, GroupID or whatever, but not SampleID, Sample-ID or other other names reserved for index.

[yangyue]

@anna.abbey

case sensitivity

I usually thought it is a case sensitivity problem, open the *.tsv file with Microsoft Excel, then save it with comma in *.csv UTF-8 format file, finally change the suffix as ＂tsv＂.

[anna.abbey]

Hi Colin,
thanks for the swift response! I was hoping to get the abundance plot for each sample at a specific time(season). Thanks for your help

[anna.abbey]

Thanks for your input. I tried this but qiime wasnt able to read my metadata file. Thanks I got it sorted.

[anna.abbey]

Hi Timur, Thanks for your input. it was really helpful and worked out!

[yangyue]

Hello Anna,

@anna.abbey

There was a problem with the command:
(1/1) Invalid value for '--m-metadata-file': There was an issue with
retrieving column 'sample-id' from the metadata.

@anna.abbey

I am wondering if this could be due to a formatting issue I am missing (e.g., hidden characters, header issues, or case sensitivity

Make sure your metadata's sample-ids actually contained in your input table "dada2_table_filt_contam_unclass_removed.qza" and your seasonal grouping column had been set correctly by your order, accompany with your sample-ids.

@anna.abbey

I was hoping to get the abundance plot for each sample at a specific time(season).

So, as you say, did you want to sort your samples with seasons? According your reply you just want to show a bar plot within a specific time (may just one season), so I think your metadata file "Abbeyv4v5_sample_1metadata.tsv" should be cotain all of the information, then the choice of --m-metadata-column should be ignored, and add this order:

--p-where "[season]='spring'"` (for example one of your time point choice is "spring" in your metadata file, the metadata file contains a column named "season")

Otherwise, I provide you another simple but not effective way to conduct your next steps (for example):

qiime demux filter-samples \
  --i-demux your_demuxed_file.qza \
  --m-metadata-file manifest_pick_your_season_spring.tsv \
  --o-filtered-demux demuxed_partial_season_spring.qza

The file manifest_pick_your_season_spring.tsv is presented as simple as below:

As you know, the file your_demuxed_file.qza must contains the listed three samples or more samples (belong to other seasons or the whole samples could be sorted by your other research purposes). After doing this, your demuxed_partial_season_spring.qza should be denoised and then you can make bar plot or other analysis. So this step "qiime feature-table group" be ignored.