I am checking some ribotypes obtained using 16S high-throughput sequencing (V4 region) of a microbial community. I have found a ribotype (obtained from clustering of sequences at 97% similarity) that is relevant to our analysis and similar at approximately 90% with species of two genera (Desulfotomaculum and Nocardia ) belonging to two different phyla (Firmicutes and Actinobacteria) (same results obtained both on RDP database and on Greengenes). It is not a chimera, we have checked it. Does someone have suggestions? Is it correct to consider it as an “unknown bacterium”? Or it is better to consider this ribotype as belonging to one of these two genera?
That will provide similar results. ARB / SILVA phylogeny are based on full lenght alignment and not just tiny bit of it. Even at this resolution, species and genes phylogeny is not always coherent (i.e. : not all genes are under the same evolutionnary trajectories).
In this respect, if you're not confident annotating a short-read, let it join the cohort of "Unassigned reads" not "Unknown bacterium" that is a bit over-confident in stating that this piece of DNA is not in any database.
In this very case, your DNA sequence is in databases: it provides significant local alilgnment. Yet, the phylogeny (and hence, taxonomy, not the other way around) can not be deciphered from this small ammount of substitutions (ie. just comparing V4).
Ribotypes, OTU, etc, are not coherent biological units and but "technical" (i.e: based on methodology and not theory) units that can be assumed to be linked to phylogeny. You have a case where this assumption is obviously not true.
Just a bit add-up:
I guess you chose V4 region because V4 provides estimates similar to those obtained with the nearly full-length fragment（AEM（2009） 75, 5227–5236 .）.
Since this OTU matters, it's not harmful and worthy to dig a little bit more of its phylogenetic affiliation. To computer scientists, this OTU might just be a short sequence, but biologist always want to know the phylogenetic information. And some reviewers are definitely gonna rise questions if you just put this important 'unknown' OTU out there without any explanation.
According to ARB manual, short sequences are going to be directly inserted into existing phylogenetic trees, i.e. 'Quick_add parsimony tool of ARB'. And this was exactly people did when they applied DGGE to study microbial diversity back to 2008. I attached the paper. Please take a look at page 309.
Another way to find more phylogenetic information about this OTU is to design a specific primer for it and couple with a universal primer to amplify the longer fragments of this OTU, which is called PCR-mediated extension of V4 tags. For instance, you can align it with other OTUs/sequences to find a unique site and design a primer. Then combine it with 27f or 1492r to amplify other hypervariable regions. This strategy has been used to study sponge symbionts. Please refer to the paper in 2010.
An in silico approach to test wether this is a frequent situation with V4, you could try to build a phylogenetic tree based only on V4 and check its coherence with a full-lenght 16S tree calculated with aligned to reference sequence (i.e.: better accuracy in nucleotide position) taken from SILVA database for example.
On a side note, and while not a phylogeny specialist but a microbial ecologist, I however do not think that any valid criteria are defining phylogentic ranks thresholds (ie. what is a new genus/order/phylum/etc). And sequence similarity alone, especially with such short sequence, are likely to be inappropriate to define new taxonomic ranks, for a variety of reasons.
That's a very old debate between taxonomy (naming) and phylogeny (relationship), but there is only one that is based on biology theory (phylogeny).
To get back to the original question:
" Is it correct to consider it as an “unknown bacterium”? Or it is better to consider this ribotype as belonging to one of these two genera? "
Honestly, I think that anyone giving you a blunt answer is most likely wrong; I mean that this is an open science debate on what can you infer from amplicons sequencing. My suggestion, however, would be to introduce this question in your results and consider both answers, seeing how much it is modifying your conclusion.
@Shanky Bhat
Dear Shanky Bhat, after a very detailed analysis on our 16S sequence we did not resolved our problem. Our sequence was the result of a clustering at 97% similarity and it was assigned to the genus Salinispora by MGRAST. We have carefully checked the result as you have done with blast, we have also used Green genes database but our sequence remained only distantly related to Desulfotomaculum and Nocardia. We are sure that the sequence do not belong to the Salinispora genus. We also tried the blast to tree software but in our hands it did not work. We have finally published the paper but, obviously, we did not reported the final annotation of the sequence. We are now doing a more detailed check using shotgun metagenomics.
Good luck!
- Badges
- Science method