Problem with vizualization of SSU in bar plot

[jko]

Hello,
I have a problem with silva-138.1-ssu-nr99-classifier.qza

Finally more RAM in machine let me install this silva, but I have new complications with them.

I get data in tables and bar plots in only form: d_Bacteria, d_Eukaryota and a lot as "Unassigned". I used primer SSU (SSUF04 GCTTGTCTCAAAGATTAAGCC , SSURmod CCTGCTGCCTTCCTTRGA) for data from aquatic ecosystem.

I have not this problem with other primer - EUK ( 1391F 5′-GTACACACCGCCCGTC-3′ , Euk B 5′-TGATCCTTCTGCAGGTTCACCTAC-3′ ). I set "Level 6" in Taxonomy Level on qiime2view in both cases.

Maybe someone has any experience in this problem?
I am quite novice in the domain.
Thank you in advance for your help!

Best regars,
jko

[colinbrislawn (Colin J Brislawn)]

Hello jko,

Let's start here. Did you make this classifier or download it from our website?
silva-138.1-ssu-nr99-classifier.qza

Can you share a citation for those two primers?

[jko]

Hello colinbrislawn,

I used tutorial by SoilRotifer to processing, filtering and evaluation the SILVA database with Rescript from this forum: https://forum.qiime2.org/t/processing-filtering-and-evaluating-the-silva-database-and-other-reference-sequence-data-with-rescript/15494

I got SILVA data by Hard Mode from this tutorial.

Citations:

1.  Forster, D., Filker, S., Kochems, R., Breiner, H. W., Cordier, T., Pawlowski, J., & Stoeck, T. (2019). A comparison of different ciliate metabarcode genes as bioindicators for environmental impact assessments of salmon aquaculture. Journal of Eukaryotic Microbiology, 66(2), 294-308.
   

1391F 5′-GTACACACCGCCCGTC-3′
Euk B 5′-TGATCCTTCTGCAGGTTCACCTAC-3′

2.  Günther, B., Jourdain, E., Rubincam, L., Karoliussen, R., Cox, S. L., & Arnaud Haond, S. (2022). Feces DNA analyses track the rehabilitation of a free-ranging beluga whale. Scientific Reports, 12(1), 1-7.
   

SSUF04 GCTTGTCTCAAAGATTAAGCC
SSURmod CCTGCTGCCTTCCTTRGA

[colinbrislawn (Colin J Brislawn)]

Hello @jok

I've been discussing this question offline with @SoilRotifer, and we think the issue may be related to read orientation.

During Illumina sequencing, the amplicon primers can be used as sequencing primers to that all the reads are in the 'forward' orientation. So R1 is always SSUF04 and R2 is always SSURmod.

Or, the normal Illumina primers can be used, so R1 is a mix of SSUF04 and SSURmod. Half the reads are 'backwards' after sequencing, and can lead to many unassigned reads.

Do you think that could be going on here?

If you want to try this, run the rescript orient-seqs option from the SILVA tutorial. If that makes your ASVs get taxonomy, that's a good step forward!

[jko]

Thank you for your answer!

I tried do it with tutorial rescript orient-seqs
https://docs.qiime2.org/2024.10/plugins/available/rescript/orient-seqs/

I used:

rescript orient-seqs
--sequences FeatureData[Sequence]
--o-oriented-seqs FeatureData[Sequence]
--o-unmatched-seqs FeatureData[Sequence]*

and then the barplot vizualization looks the same, only d_Bacteria, d_Eukaryota and "Unassigned"

.
.
.

qiime rescript orient-seqs
--i-sequences rep-seqs_ssu.qza
--o-oriented-seqs table_ssu_oriented.qza
--o-unmatched-seqs table_ssu_unmatched.qza

qiime feature-classifier classify-sklearn
--i-classifier silva-138.1-ssu-nr99-classifier.qza
--i-reads table_ssu_oriented.qza
--o-classification taxonomy_ssu.qza

qiime metadata tabulate
--m-input-file taxonomy_ssu.qza
--o-visualization taxonomy_ssu.qzv

qiime metadata tabulate
--m-input-file taxonomy_ssu.qza
--m-input-file table_ssu_oriented.qza
--o-visualization annotated_rep-set_ssu.qzv

qiime taxa barplot
--i-table table_ssu.qza
--i-taxonomy taxonomy_ssu.qza
--m-metadata-file metadata_algssu.txt
--o-visualization taxa-bar-plots_ssu.qzv

[colinbrislawn (Colin J Brislawn)]

Would you be willing to post the full taxa_barplot.qzv file so we can take a look?

[jko]

Hello,
Off course. It is here:

taxa-bar-plots_ssu.qzv (384.1 KB)

[SoilRotifer (Mike Robeson)]

Hi @jko,

When you re-run feature-classifier classify-sklearn ... , can you set the option --p-read-orientation same. Sometimes the default --p-read-orientation auto can be tricked and result in spurious results, so your command should be:

qiime feature-classifier classify-sklearn \
    --i-classifier silva-138.1-ssu-nr99-classifier.qza \
    --i-reads seqs_ssu_oriented.qza \
    --p-read-orientation same \
    --o-classification taxonomy_ssu.qza \
    --verbose

I noticed you mislabeled your output as table_ssu_oriented.qza, when it should probably be repseqs_ssu_oriented.qza.

Anyway, if setting the orientation does not help... then I'd try running the following using the reference sequences and taxonomy files (those used to make the classifier) as input:

qiime feature-classifier classify-consensus-vsearch \
    --i-query repseqs_ssu_oriented \
    --i-reference-reads silva-138-99-seqs.qza \
    --i-reference-taxonomy silva-138-99-tax.qza\
    --o-classification repseqs_ssu_oriented_taxonomy.qza \
   --o-search-results  repseqs_ssu_oriented_top_hits.qza \
   --verbose

Vsearch will work regardless of orientation. But if you obtain the same spurious results... then I would think the sequencing run did not work as intended. Probably manually BLAST a few of the sequences online with the megablast setting, and excluding environmental sequences.