SEQanswers

SEQanswers (http://seqanswers.com/forums/index.php)
-   Metagenomics (http://seqanswers.com/forums/forumdisplay.php?f=29)
-   -   16S triplicates (http://seqanswers.com/forums/showthread.php?t=70837)

lorendarith 08-12-2016 09:09 AM

16S triplicates
 
I often see in studies that whenever people prepare samples for 16S sequencing there's often a line which says that each sample was amplified in triplicates.
Why?

Is there any paper out there which explains the need for having 16S samples in triplicates? I couldn't find anything informative...

I understand why somebody would make triplicates in cases where the amplification for that sample would fail due to technical issues. You'd end up with less samples and have to repeat some maybe, but from an analysis point of view, is there really a need?
Is it to potentially average out the variation caused during amplification or sequencing? Any papers showing this effect?

It's not an issue if your sample size is small, but if you are dealing with 1000s of samples, it does get quite costly in terms of PCR/adapter oligo synthesis and PCR mix quantities.
Just trying to avoid unnecessary cost...

thermophile 08-12-2016 12:14 PM

You should amplify your sample more than once because the seqs that get amplified in the first few rounds of amplification will dominate the amplicion pool. Since this is random, more common seqs should be among the first amplified but sometimes they aren't. The more reactions you initialize the more often you should see the "true" most abundant seqs being the most abundant at the end of PCR.

To minimize costs, you can reduce reaction volumes.

Google "16s triplicate pcr" for lots of papers to cite.

*"true" is a very lofty goal for 16S surveys but lets just go with it)

lorendarith 08-15-2016 04:01 AM

Quote:

Originally Posted by thermophile (Post 197780)
You should amplify your sample more than once because the seqs that get amplified in the first few rounds of amplification will dominate the amplicion pool. Since this is random, more common seqs should be among the first amplified but sometimes they aren't. The more reactions you initialize the more often you should see the "true" most abundant seqs being the most abundant at the end of PCR.

To minimize costs, you can reduce reaction volumes.

Google "16s triplicate pcr" for lots of papers to cite.

*"true" is a very lofty goal for 16S surveys but lets just go with it)

Thanks for the answer. I am not looking for papers to cite, but for papers which speak for using more than one PCR reaction for 16S sequencing.

If this particular case of PCR template bias is a thing, how come just three reactions would be enough to solve this issue? This would be heavily dependent on the complexity of the template DNA. I would imagine that in this case you'd need way more PCR reactions for soil samples than for human gut samples as an example. Yet studies on both settle for just 3 replicate reactions.

There are countless 16S seq protocols which use triplicate reactions yet almost none of them explain why, not even if a single reference.

The only paper I found which tries to look at triplicate vs. single reactions:
Evaluating Bias of Illumina-Based Bacterial 16S rRNA Gene Profiles
Katherine Kennedy, Michael W. Hall, Michael D. J. Lynch, Gabriel Moreno-Hagelsieb, and Josh D. Neufeld
Appl. Environ. Microbiol. September 2014 80:18 5717-572
2
says:
Quote:

Pooling of triplicate amplifications was first proposed to decrease PCR bias introduced by stochastic fluctuations in amplification efficiencies (drift) (3). However, the impact of pooling has not been demonstrated previously.
which I found is incredibly mindblowing that researchers were following this triplicate pooling practice for years without any evidence that it really helps??? Because this paper only indicates that it is very situation and could help mostly with very similar samples, to sort of reduce noise in the data.
Continuing:
Quote:

Although one would predict that pooling PCR amplifications might contribute distinct representation from low-abundance “rare biosphere” organisms, we did not detect any significant differences in key alpha diversity metrics (e.g., Shannon, Chao1, and numbers of OTUs) between pooled and individual amplicon data for each sample (data not shown).
As a reason for pooling, this study cites a paper from 1998 which highlights template bias in a simple DNA mix yet actually doesn't show anything on replicating PCR reactions:
Bias in Template-to-Product Ratios in Multitemplate PCR
Martin F. Polz and Colleen M. Cavanaugh
Appl. Environ. Microbiol. October 1998


I'm just baffled at the lack of evidence for this practice. Figured that by now somebody could have at least taken different mock community (increasing in complexity) and amplified it in several reactions and different numbers of pools to demonstrate this particular effect, given the advance in sequencing as well as computational methods for analyses we just witnessed. Multiple companies boast about these new polymerases with decreased error rates and biases. I would find it very weird all these advances would not impact this particular pooling practice.

adam.geber 10-11-2016 08:27 AM

I think this may be one of the papers you're seeking out: Sequence Depth, Not PCR Replication, Improves Ecological Inference from Next Generation DNA Sequencing

The authors assess the relationship between pooled PCR replicates and diversity metrics for 16S profiling on both the 454 and the MiSeq and show little benefit to replicate PCR.


All times are GMT -8. The time now is 07:38 AM.

Powered by vBulletin® Version 3.8.9
Copyright ©2000 - 2020, vBulletin Solutions, Inc.