Sequences disappeared after running qiime feature-table merge-seqs

[Martin_Kostovcik]

Hi Team,
I'm experiencing similar/same issue as mentioned in topic : sequences disappear with this command: qiime feature-table merge-seqs
which wasn't resolved. After merging several representative_sequences.qza artifacts it turned out that 2 of the sequences disappeared in the merged file - the features are present in merged table artifact and the way I discovered it was after getting error message when trying to run qiime taxa barplot fot the merged feature table with taxonomy created from merged representative_sequences.qza artifact.
My commands were:

qiime feature-table merge --i-tables /media/storage/Galanova_0520/H4/dada2out/table.qza   /media/storage/Galanova_0520/H5/dada2out/table.qza  /media/storage/Galanova_0520/H6/dada2out/table.qza   --o-merged-table /media/storage/Galanova_0520/Fungi_joined/merged_table  --p-overlap-method sum;
qiime feature-table summarize --i-table /media/storage/Galanova_0520/Fungi_joined/merged_table.qza  --m-sample-metadata-file /media/storage/Galanova_0520/Fungi_all_mapping.txt  --o-visualization /media/storage/Galanova_0520/Fungi_joined/merged_table;
qiime feature-table merge-seqs --i-data  /media/storage/Galanova_0520/H4/dada2out/representative_sequences_H4.qza   /media/storage/Galanova_0520/H5/dada2out/representative_sequences_H5.qza   /media/storage/Galanova_0520/H6/dada2out/representative_sequences_H6.qza    --o-merged-data  /media/storage/Galanova_0520/Fungi_joined/rep-seqs-dada2_full_merged;
qiime feature-table tabulate-seqs --i-data  /media/storage/Galanova_0520/Fungi_joined/rep-seqs-dada2_full_merged.qza  --o-visualization /media/storage/Galanova_0520/Fungi_joined/rep-seqs-dada2_full_merged;

I'm running qiime2-2021.4 nativelly installed.
Thanks for any advice.
Martin

[thermokarst]

Hi @Martin_Kostovcik - can you please share the two summary QZVs you listed above:

1. merged_table.qzv
2. rep-seqs-dada2_full_merged.qzv

[Martin_Kostovcik]

@Martin_Kostovcik

/media/storage/Galanova_0520/Fungi_joined/merged_table

Hi,
here it is. Thanks for looking into it and eny clues about the issue.
Martinrep-seqs-dada2_full_merged[1].qzv (523.0 KB) merged_table[1].qzv (572.2 KB)

[thermokarst]

Thanks for sharing those files @Martin_Kostovcik!

@Martin_Kostovcik

it turned out that 2 of the sequences disappeared in the merged file

Can you elaborate further on why you think you're missing features? The two visualizations you shared indicate that the feature count is matched between the two:

merged seqs:

merged tables:

[Martin_Kostovcik]

Oooh,
my fault. Sorry I was a bit confused with the identical name of the file and uploaded a different on. So once again and better attaching two rep_seq files which after merging lost 2 features - fc57242f57cf161639b4180ff801742b and 355f5cde4f0b2ac57f32bb419ff55c33. I found out that after getting error from classification abous missing features. Those features are present in the first rep_seq files being merged.....no idea what happened. Sorry once again for attaching incorrect file i was in hurry a bit.
The command was : qiime feature-table merge-seqs --i-data /media/storage/Galanova_0520/Fungi_joined/rep-seqs-dada2_full_merged.qza /media/storage/Kostovcikova_bobky_0521/210510_Kostovcikova20210510/raw_fastq/Fungi/KostH4/dada2out/representative_sequences.qza --o-merged-data /media/storage/Kostovcikova_bobky_0521/210510_Kostovcikova20210510/raw_fastq/Fungi/TD_KK_OLOMOUC_FUNG_JOINED_W_0520/rep-seqs-dada2_full_merged;
rep-seqs-dada2_full_merged[1].qzv (523.0 KB) rep-seqs-dada2_full_merged[1].qzv (683.3 KB)
The second one is the merged file.
Thanks for any clue.
Martin

[thermokarst]

Hi @Martin_Kostovcik - I suggest you double-check your book-keeping - these two visualizations have completely different provenance. If you trace back, even the import steps show that you imported different data between the two runs. Double-check that you have everything in order and let us know what you find.

[Martin_Kostovcik]

Hi Matt,
so I double-checked the files renamed the outputs and it's working now by what I mean those two missing features are in the merged table as well as in merged rep_seq file. I used exactly the same command as in a previous post just changed the name of the output file. I'm attaching once again both files - the old one with missing features (fc57242f57cf161639b4180ff801742b; 355f5cde4f0b2ac57f32bb419ff55c33) and the new one without missing features. Both files as I mentioned, were created using the same command : qiime feature-table merge-seqs --i-data /media/storage/Galanova_0520/Fungi_joined/rep-seqs-dada2_full_merged.qza /media/storage/Kostovcikova_bobky_0521/210510_Kostovcikova20210510/raw_fastq/Fungi/KostH4/dada2out/representative_sequences.qza --o-merged-data /media/storage/Kostovcikova_bobky_0521/210510_Kostovcikova20210510/raw_fastq/Fungi/TD_KK_OLOMOUC_FUNG_JOINED_W_0520/rep-seqs-dada2_full_merged_test; There is a difference of 5 extra features in the latter "test" file.rep-seqs-dada2_full_merged[1].qzv (683.3 KB) rep-seqs-dada2_full_merged_test[1].qzv (686.5 KB)
No idea where was a problem but it's working now. Thanks for your help and guidance.

[thermokarst]

@Martin_Kostovcik

No idea where was a problem but it's working now

Hi @Martin_Kostovcik - the initial data import represented completely different data (I'll include just the md5 sums for the barcode files imported, for brevity)

• viz #1 (UUID 193291cf-6b93-4f63-82bd-72d11f47f632):
  • 7e4716b1a417ab41344a7c30dd4dd5df
  • 9dca27c5264f9651a1f6be5dc3b2692a
  • a84ed9453e5a263ef12da9fc8de1b37c
• viz #2 (UUID 416e832d-c25b-4b50-8645-61af65d31f3a) (note, this file was shared twice in this thread):
  • 26f10d401ea1f3d1791b5b1a6e513775
  • 442d55f690a8ce50fd91a23c98a37595
  • 8d4a528bedb4426a364d5de895bbb0d1
  • f0c2f5321f4451ffd3efa29bf425836c
• viz #3 (UUID c5d9035f-5774-4970-976c-8f5074154880)
  • 7e4716b1a417ab41344a7c30dd4dd5df
  • 8d4a528bedb4426a364d5de895bbb0d1
  • 9dca27c5264f9651a1f6be5dc3b2692a
  • a84ed9453e5a263ef12da9fc8de1b37c

This is why I'm thinking you have a book-keeping issue - it sounds like you got everything sorted out, but in case something like this happens again, pay careful attention to what files you are using, and where/when. Good luck!

PS - you can see those md5sums here when you click on the "provenance" tab on view.qiime2.org after loading a file: