Losing samples when merging DADA2 tables

[zzc (Zechariah Carroll)]

Hi there!
I am having difficulty merging feature tables generated by DADA2.
I have four feature-table artifacts comprised in total of about 940 samples, but only 482 of them pass to the output table. Three of the tables pass their samples without issue, yet none of the samples from the fourth file make it to the final table.

Here is the command I submitted:

qiime feature-table merge
--i-tables table-dada2_1.qza
--i-tables table-dada2_2.qza
--i-tables table-dada2_3.qza
--i-tables table-dada2_4.qza
--p-overlap-method sum
--o-merged-table merged-feature-table.qza

I am unable to upload the erroneous table (table-dada2_4.qza) due to size, but would greatly appreciate any assistance in troubleshooting this problem.
Thanks a million!

[colinbrislawn (Colin J Brislawn)]

Hello Zech,

I would like to know the answer to this question too. I know that merging DADA2 tables is possible (unlike older de novo OTU tables), so I look forward to hearing the recommendation of @Nicholas_Bokulich or another DADA2 dev.

Colin

[ebolyen (Evan Bolyen)]

Hi @zzc,

@zzc

I have four feature-table artifacts comprised in total of about 940 samples, but only 482 of them pass to the output table. Three of the tables pass their samples without issue, yet none of the samples from the fourth file make it to the final table.

That is very peculiar. Could you provide a feature-table summary of both table-dada2_4.qza and merged-feature-table.qza? Those visualizations should be small-ish and I can double check the provenance to make sure what we think is happening is happening. (No need to pass --m-metadata-file into your feature-table summarize command either.)

Otherwise, everything looks right to me, so I don't have any easy explanations yet.

Thanks!