Quantitative pyrosequencing?

[skolyar]

Dear all,
I am new to NGS and need some advise please.
Would it be sensible to use deep sequencing of PCR amplicons (e.g. 16S gene in genomic DNA) for quantitative studies of complex microbial population in a number of samples? Will this sequence informantion retain the initial proportional composition of the populations and can it be used for a comparative quantification of population composition between the samples? If so, how many PCR cycles should be used for the amplification to avoid bias?
I hope it all makes sense..
Thanks

[pmiguel]

This is a perennial issue. For each new technology the same questions emerge. But this need not be the case. Any lab course in biology will include the basics of experimental design and frequently will address post-experimental analysis of results. No new technology, no matter how flashy, circumvents the concerns that good experimental design is meant to address.

In brief, the answer is two-fold.

(1) If you are doing a "descriptive" experiment, then you might forgo replicates and spike-in controls for depth-of-coverage and speed of data acquisition. You can usually go back and validate results that strike you as interesting using another method.

(2) Otherwise, all the normal practices of experimental design are applicable.

One further caveat is "publishability". Can the results you obtain be published? Generally, as time passes, standards and practices for an assay develop in groups of peer reviewers. Some of them even end up formalized in guidelines for submission to a journal. Frequently new methods are given more leeway in the literature. Their novelty will trump lack of rigor.

Why, you may ask, don't I just answer your questions? It is because I don't have an answer. I don't think anyone does. I could speculate: no, a PCR amplification of rRNA genes will not faithfully represent the make up of a complex microbial population. How could it? The sequence of the primers you use will bias your results. The number of copies of rRNA genes per genome will bias your results. That happens even before you get into PCR, emPCR and sequencing.

But, on the other hand, will you know more than if you did not do the experiment at all? Probably.

But, hey, this is all good news. It means that the decisions you make are important and will likely determine whether the data produced will be of value or not. Better than being a small cog in a big machine, eh?

--
Phillip

[skolyar]

Dear Phillip,
Thanks for your comments and sharing your views on this. Well, it looks like it all comes down to “suck it and see” type of experiment.