We can easily get tons of 16S rRNA sequence data for microbial communities by next-gen sequencing techniques with primers (eg., V3 to V4). If we want to compare whether bacterial communities are different among samples, at what taxonomic level should we compare?
Some compare bacterial communities at phylum level, some at class or genus level. What is the rationale for those comparison? At what level do you suggest to use? Some communities might be similar at phylum level but may not at genus level. At genus level, however, more sequences become unclassified due to lack of reference.
The answer depends on why you are comparing these communities.
In our case, we prefer genus level . Although we cannot always identify them , most frequently we are interested in specific bacteria. Regarding the concern on assignment we do not see a significant proportion of sequences unclassified. Finally, the comparison can be done for OTUs . Each of OTUs can be assigned to as precise as possible. Thus , you do not need to decide in advance which level of taxonomic assignment is acceptable.
Hi,
I'm not pretty sure if you are assessing only amplicon 16S or metagenome approach, they are completely different approaches that require be analyzed in distinct ways. Given your information, I assume you are studying amplicon 16S, and in this case you are limited by the regions you included in the amplicons sequenced. I'm afraid that unless you won't sequence fragments larger than 1kb, the taxonomic assignation even a genus level will remain elusive. If you are using two hypervariable regions (V3-V4, or V4-V5), I recommend you get taxonomy at family level that will leave you with a high level of certainty over such identifications.
Hi Daniel
Im reinforcing the talk of previous colleagues. Unfortunely, resolution based on amplification of V3-V4 16S region is not the best way to determine the species with high level of confidence. However, it provides a useful tool to characterize microbial comunity structure based on higher taxonomic level such as phyla genus or family. To determine species level with high level of confidence is more apropriate to use larger fragments. But even in this way, you can establish relation between some functional ecological roles based in those taxonomic levels.
Thank you all. How do you deal with the unclassified sequences? The proportion is around 5% to 10%, sometimes even higher (20%) at genus level. Just delete them or treat them as one as one column as other OTUs, or some other solutions? How do you analyze those data? Any suggestions?
It´s depend on your question. If you are considering biological diversity you can use up to family or genus. For example, 60% Pseudomonadaceae, 30% Enterobacteriaceae and so on. Within these percentages, you can describe genus up to complete 60%, 30% and finish.
So, as I wrote before... It´s depend on your question. What software are using? Geneios, Mega, Sequencher?
Kinds,
C
I'm just getting into this area myself, so forgive me (and correct me) if my comments are naive and/or incorrect.
It appears that the poor accuracy of HTS limits the taxonomic resolution obtainable. Rather than generating a single accurate read and hand editing out ridiculous calls as in Sanger Sequencing, HTS works by averaging over a large number of much more error-prone reads. That means you need an enormous number of reads of the same sequence to generate a 100% certain consensus, if indeed it is possible at all.
Logic says that the most abundant sequences should give greatest sequence depth and therefore greatest taxonomic accuracy, while poorly represented sequences will be much less accurate.
Some quick thoughts: (1) taxonomic ranks are artificial and not really comparable, although they could provide a sense of similarity and difference; (2) unassigned sequences may be due to lack of reference sequences in the database, which needs to be manually modified (I had to add a few symbiont sequences not represented in the GreenGene database); and (3) distance-based OTU clustering and taxonomic assignment is inherently problematic as it is not phylogenetic and is at bast a proxy to phylogenetic entities. Take-home message: bacterial diversity research has to move towards a phylogeny-based paradigm, which is currently lacking. So the task is upon us to change it.
- Badges
- Science topic
- Similar topics
- Biological Science
- Molecular Biology
- Biomolecules
- Nucleic Acids
- RNA