Archive: The ‘Unified Microbiome Initiative’ and the Risks of Standardization

By: Noah Fierer

October 29, 2015

Recently there was a call from a group of esteemed microbiologists and microbial ecologists for a ‘Unified Microbiome Intiative’. I won’t go into the details as you can read the articles here and here.

I agree with most of what the authors are proposing. However, there are a few aspects of this proposed intiative that left me a bit grouchy. Fortunately, I was not the only one with some concerns as Pat Schloss has already written an excellent blog post with his reactions to the proposed Unified Microbiome Initiative. He beat me to the punch and most of what I wanted to add to the conversation was already highlighted by Pat. Nevertheless, here are some cursory thoughts on the proposed intiative:

First, I think it is great to have more funding for microbiome science. Take this with a massive grain of salt as I run a lab that is mostly funded to study microbes in air, soil, plants, and other environments so it would be stupid for me to argue against more funding for this discipline. Of course - I am also biased by my inordinate fondness for microbes, but I do think microbiome research is worthy of funding as it is one of the more interesting and rapidly growing fields of biology at the moment. Yes – there is far too much hype and microbes are not the answer to every conundrum in biology. Comments such as this one:

“any attempt to understand anything about plants or animals needs to be rooted in microbiology”

will never endear microbiologists to ‘macro’-biologists and such statements are demonstrably untrue – we can’t ignore hundreds of years of research on plants and animals just because prior studies might not have looked at their microbiomes. Nevertheless, I remain amazed by the preponderance of unanswered, important questions about microbes and their effects on plant/animal/ecosystem health. We now have methods suitable for addressing some of these big questions and microbial ecology is clearly in a ‘golden age’ (though probably every scientist thinks their field of choice is in a ‘golden age’ - just once, I’d love to hear a scientist say out loud “Yep – my field is past its prime, it is no longer very exciting and all the important discoveries have already been made”).

One key feature of the ‘Unified Microbiome Initiative’ is a call for greater standardization of methods across research groups. I do not think this is the way to advance the science. Yes – having research groups across the globe use a more standardized set of methods will improve the ability to compare datasets across studies. Such standardization may also make it easier for newcomers to jump into the field. For example - instead of wading through hundreds of papers describing the ‘best way’ to do marker gene analyses, a lab would be able to identify the method of choice (or at least the method identified by a select group of scientists as being ‘the best’) and proceed full steam ahead. However, promoting widespread use of a standard set of ‘consensus’ methods also comes with inherent risks.

Why try to promote standardization when we know the methods we are using to study microbial communities are rapidly advancing and no current methods are ideal?

There are a number of downsides to pushing for more standardization of methods by microbial ecologists. First, it is highly likely that the chosen consensus method may not actually be very good and may have important limitations – limitations that may not be readily apparent. For example, if all research groups had decided to use the PCR primers chosen by the Earth Microbiome Project for 16S rRNA analyses of bacterial communities, we would all be using primer pairs biased against many bacteria - including a dominant skin bacterial genus, Propionibacterium, and one of the most dominant bacterial groups on the planet – the marine SAR11 clade. Likewise, many commonly-used PCR primers miss an entire bacterial phylum that dominates many soils. We might never have known that members of this group are so ubiquitous and abundant if we had all decided to use the same ‘consensus’ primers.

Standardization rarely promotes innovation. For example, there are dozens of bioinformatic strategies for clustering 16S rRNA gene sequences into operational taxonomic units (OTUs) and all of them have their limitations (e.g. computational efficiency, accuracy, resolution, over- or under-estimation of diversity, etc.). To decide that one clustering algorithm is, by consensus, ‘the best’ would stifle innovation in this critical aspect of sequence data analysis. It is very clear that there is no single ‘best’ algorithm – the choice will depend on your computational resources, your research questions, and the sequence data available. Plus, methods are constantly changing (hopefully improving) and many of these improvements come from smaller labs that may not be directly involved in the consortium that decides the consensus methods.

Note I’m not suggesting that we go about ‘business as usual’. There is a lot of utility in testing methods rigorously, making datasets publicly available that can effectively be used for meta-analyses, and coordinating activities across research teams, funding agencies, and disciplines. However, there are also some advantages to investigator-driven research and I would hope that the Unified Microbiome Initiative would never (directly or indirectly) lead to homogenization of scientific efforts. It would be a shame if microbiome researchers were encouraged to all travel down the same well-trodden path, a path that may not be the most efficient route to scientific advancement.