MiSeq V4 16S run low OTU evenness

[Egansbay]

Hi everyone

Recently received my sequences back from an Illumina MiSeq 16S V4 300bp PE run, performed by a biotech company. The DNA was extracted from the gut of fish.

Although I received many millions of reads (approx. 1m raw reads/sample), after running the samples through Mothur, there seems to be a dominance of one particular otu, up to 99% in the case of some samples.

I am wondering whether anybody else experiences similar patterns? I know each study is different depending on the environment, but I am just wondering if the pattern of one otu totally dominating is a common one with these types of NGS studies?

Thank you

Egansbay

[gblanchard4]

I haven't encountered that in any of the sequencing my lab has done. Have you tried any other clustering methods? What percentage are you clustering at? How many total OTUs did you find?

[Egansbay]

I set the cutoff=80 on the classify.seqs command. About 500-800 otu's total. Not sure if I've done something wrong or whether this is a true representation

[cliffbeall]

It might be possible, we have seen dominance by one species in vaginal samples and some oral samples from small children.

About generating millions of sequences per sample, though- I'm not sure this is really productive. We have measured 16S copy numbers in various metagenomic DNA samples and they are often in the hundreds of thousands. So you may be spending money sequencing PCR duplicates.

[Egansbay]

Hi Cliff - yes I have found this to indeed be the case when I have sequenced further sample sets.

How many v4 reads per sample are you generally using for classification, after quality filtering?

I am finding that rarefaction reaches full saturation after approx 150,000 reads/sample for the majority, but this is for the fish gut

Really appreciate your help

Cheers

[AlexT]

can you comment on the otu you found?
we have seen such effects also in the gut of fish but thought that this is more a kind of contamination due to insufficient input dna

[Egansbay]

Hi Alex, appears to be a Mycoplasma. Appearing a lot in the fish gut microbiome literature but also a contaminant of cell culture. Same otu?

[AlexT]

no, we had a proteobacteria; when we tested several sites of the lab, it appeared again and again; it is now our "negative control". I would therefore propose to do this in your lab, too.

[Egansbay]

Ok I see - we tested an extraction from the reagents we are using but the otu was not present. We also tested an extraction from the diets used to feed our fish but again negative.

Are you using v4 miseq? How man reads do you use in your analysis, after quality filtering as a matter of interest?

[AlexT]

we also had an V4 fragment on a miseq, but using 2x250bp
did you get no result in your tests or a diverse population without your otu?

the number of reads we are using depends on the project of our customers, starting at 100.000x

[Egansbay]

Ok yes I have also used 2 x 250bp before too

Yes I am getting a diverse community besides the primary OTU, and anywhere from 100,000-600,000 seqs lefts over after quality filtering

So after quality filtering and prior to classification, you are getting >100,000 reads per sample? How about for your fish gut samples?

[AlexT]

no, we start at least with 100,000, usually we aim for more

[Egansbay]

Hi Alex - sorry to be a pain, but just want to clarify.

Do you mean that you have a minimum of 100,000 sequences per sample left over after all quality filtering steps? Or is this your starting number before quality filtering?

Thank you

[AlexT]

no problem, it is the starting number before quality filtering; but as I wrote before: we usually start with more reads, this is the minimum number

[Egansbay]

Ok I see - we usually get approx. 900,000 reads per sample after the sequencing run of 24 samples, but after quality filtering this drops by anywhere from 30-60%.

How many sequences do you then have per sample after the quality filtering steps?

Some papers seem to have very few, whereas some are around 300,000 or sp per sample. Difficult to know but I suppose it depends on whether the flow cell was used alone or in conjunction with other samples etc?