Salmonella, again


I am having an issue here. I have Salmonella in samples at a detectable level (assessed by culture), but they are not assigned taxomically, even though my positive control (Zymo) that contain Salmonella, has it’s Salmonella correctly assigned. Weird

Basically, if I do not consider my controls, what I beleive is happening is that my Salmonella (serovar Thyphimurium) are most probably in Enterobacteria_unclassified. Here is what I did to troubleshot the problem.

I classified before OTU clustering and before pre-cluster: still no Salmonella, so it is not a matter of the sequences being ungulfed with other unrelated sequence.

I tryed lowering kmer to 6: no change. I tried to increasing it to 10: more unclassified were assigned to Shigella and E.coli (close relative to Salmonella), but not to Salmonella, and a good chunk remain as Enterobacteriaceae_unclassified.

switched to knn for classify.seqs : worsened the classification.

Tried PDS and Silva: no difference

Lowered cutoff for boostraping down to 50%: no difference…

Used get.lineage to get sequence for Enterobacteriaceau_unclassified and blasted them. Got a bunch of Klebsiella but my first sequence, the most abuundant one, got a Salmonella hit (100% coverage, 248 identities out of 253). Then put it in Silva (online fasta default search) and it returned no match even if I get a 98.5 identity match. I have contacted Silva about this issues as well.

What do you think? Got any solution for me?

Here is the answer from Silva

Dear Alexander,

I have pasted your sequence into our online alignment and classification tool and it is classified as Salmonella. My assumption is that the classification was not active for your job and so only the alignment was done. This checkbox needs to be checked (unchecked by default):

If you take only the 16S for your Salmonella and trim it to the region you sequenced, what do you get from our classifier?

Aside from it being misclassified, I strongly suspect that while Salmonella was present in your samples, it was below your limit of detection. In my work we work with C. difficile. We can get many colonies on a plate and not see it by sequencing because even 106 CFU/g of feces is still pretty rare in the community


Hello ,thank you for the quick answer. I beleive that I have salmonella in my samples, just that it does not get assigned correctly. I do have sequences that looks like Salmonella in my get.lineage fasta file,


But anyway, I give up, tried my best but they just do not show. My last resort is to use older version of Silva, like 132, whre in past studies I could identify salmonella in in vivo infected animals, althoug it was a Salmonella Enteritidis that was used and not a thyphimurium.

Kind regards,

Well, v132 did not change a thing. Too bad. I will figure something out!


I suspect it might be below your limit of detection by sequencing.

This topic was automatically closed 10 days after the last reply. New replies are no longer allowed.