Hello everyone,
I am having some trouble implementing the dist.seqs + cluster commands for getting OTUs.
My data consist in V3-V4 amplicons (464 bp) from 48 intestine samples. I analyzed these data with QIIME 1.9 and with mothur MiSeq-SOP (using SILVA_132 database). With QIIME my final .biom table contains 1500 OTU aprox. But when I use mothur dist.seqs + cluster (with 0.03 cutoff) my final OTU table contains more than 340k OTU.
I assume I’m doing something wrong in the pipeline, maybe in previous steps.
I will appreciate if someone could give me some advice.
I suspect you’re doing something like closed or open reference clustering in qiime. You are running into a problem because the data quality for V3-V4 region will be pretty poor since the reads will have about 25 nt of overlap. I’d encourage you to look over this blog post.
Very enlightening, thank you very much.
Indeed, my QIIME pipeline included an open reference clustering, and reading your post I am wondering if:
a) mothur does not have an equivalent method for open-reference otu picking, or
b) mothur has this method but it works in a different way.
In both cases I conclude that constructing mothur and QIIME equivalent and comparable pipelines for V3-V4 data processing may be not possible, if including OTU clustering approaches… am I wrong?
I apologize if these questions are so newbie, I tried to find answers by myself but unsuccessfully.
At this point, we do not have an open-reference clustering algorithm. We have written several papers that critique the approach as implemented in QIIME and find it has many problems. You should consult these papers…
You should instead be using de novo approaches which are evaluated in the above papers as well as in this paper:
1 Like
This topic was automatically closed 10 days after the last reply. New replies are no longer allowed.