Monday, August 08, 2011

Uggh: "Barcoding" researchers keep ignoring microbes and history #antimicrobiites #annoying

Barcoding is a technique in which researchers sequence a small region of the genome of an organism and use this sequence information as a "tag" to identify species/types.   The term barcoding is most commonly associated with sequencing a small portion of the mitochondrial genome of a multicellular eukaryote.  There is a new paper in PLoS One discussing barcoding: Neotropical Bats: Estimating Species Diversity with DNA Barcodes.  The paper seems potentially interesting.  But I could not get past the part where they write "To the best of our knowledge, it is one of the largest molecular surveys of biodiversity."

Let's examine this statement a bit.  In the study here the authors did a heroic task - they did barcoding for some 9000 individuals.  It is impressive, in many ways.  But one can only call it a large molecular survey of biodiversity if one thinks  biodiversity = plants and animals. If however you include microbes, as of course you should, then this new study is not even remotely the "largest molecular survey of biodiversity."  In fact, one could argue that 100s if not 1000s of studies of biodiversity of microbes are "larger" in many ways than this study.  In microbial studies 9000 "individuals" are characterized routinely in most studies via the use of collecting DNA from environmental samples and sequencing genes from 100s -1000s to even billions of individual cells in a sample.  This is done routinely in both metagenomic work (where one collects DNA from the environment and randomly sequences it) or ribosomal RNA PCR surveys where one collects DNA and then sequences rRNA genes from the sample.  Overall, hundreds of microbial studies cover more biodiversity than this one - more species - wider phylogenetic diversity - more samples even.


So - could it be that the authors of the current paper are not aware of the microbial work?  I don't know - for they do not even mention it in this paper.  This is amazing / startling / annoying / disappointing.  After all, molecular studies of biodiversity were first done in microbial studies in the mid 1980s - pioneered by people like Norm Pace and colleagues surveying ribosomal RNA from mixed communities like Yellowstone hotsprings (e.g., see here and here). And in the 80s and 90s microbial work covered incredibly diverse ecosystems with many studies determining 100s-1000s of sequences. As sequencing got cheaper and cheaper, molecular studies of microbial biodiversity expanded to go beyond just single genes to studies of "metagenomes" from the environment. And in single gene studies, researchers now routinely sequence millions to billions of representatives of single genes using rRNA PCR and Illumina sequencing, for example.

Should these authors here be discussing microbes?  I think so.  After all, not only would this give their paper historical context.  But it would almost certainly give it scientific context and value since there are hundreds of papers on microbes looking at species richness, biodiversity metrics, and such ((a simple pubmed search found 1357 papers using rRNA PCR and diversity as the query, for example).  And since the work here is on mitochondrial DNA there may be even more parallels to microbes than one might think at first blush. 

I note - this is not the first time Paul Hebert, the senior author on this study has gotten on my nerves about barcoding in terms of not mentioning microbes. I even posted a "dissent" regarding one of his earlier papers on Faculty of 1000 (which I used to contribute to before they become non open access). I am not sure what the explanation is, but the lack of referencing the historical work on using DNA to study diversity is disappointing (though I note in at least one of the first papers by Hebert, there was more brief mention of microbes). Not mentioning the 25 years of work on molecular studies of biodiversity in microbes gives the impression that Hebert invented using DNA to study biodiversity.  And that is not needed.  Barcoding has some nice features and uses.  No need to purposefully or accidentally make it seem like it is something new.

Are microbial studies the same as the barcoding studies being done?  No.  Many barcoding studies have things like individual voucher specimens and museum collections.  And microbial studies frequently have mixed samples like soil or water and DNA.  So what.  The general point of many of these studies is the same - using molecular data to infer information about species richness, beta diversity, phylogenetic patterns, etc.  And for this, studies of microbes long preceded the barcoding approach.  And there is a lot of useful literature out there as well as tools, methods and concepts.  I for one peruse the barcoding papers to see if there is anything useful there for my work.  It would almost certainly be good for the barcoding researchers to check out the microbial literature.  And if the leaders in the barcoding arena continue to not mention microbes, well that would be unfortunate.  

12 comments:

  1. There are plenty of people within the barcoding world who work on microbes, including a working group on protists, others on fungi, and there is environmental "barcoding" of bacteria using next-gen sequencing.

    Seems like you're latching on to one (poorly worded) sentence here.

    ReplyDelete
  2. Really? There's not a single mention of microbes in the paper yet the paper is about estimating richness and such things which have been done w/ microbes for years.

    ReplyDelete
  3. Fair enough. That paper seems like it should have been revised during review to include mention of that. (It was a PLoS journal, no? Who edits/chooses reviewers?). But as I said, it doesn't reflect the real work being done even within the barcoding initiative, contra your criticism of the entire project.

    ReplyDelete
  4. I like some papers on barcoding but overall the field seems to ignore the work already done on rRNA surveys of microbes.

    I note - as an aside - in my first class I taught at UC Davis students had to pick a paper to present and discuss with the class. One of them presented this paper: http://www.erin.utoronto.ca/~w3bio/bio443/seminar_papers/promise_of_dna_barcoding.pdf. By you and Paul Hebert. And they all asked "Why is there no mention of microbes in this paper?" I said I had no idea but I would look into it. And every time I look into it it seems to me that the barcoding field does not give enough credit to Pace, Sogin, and the entire history of work on rRNA studies and now metagenomics. Sure now some people talk about barcoding fungi and microbes. But really, how is this different from rRNA surveys? And why does it need to be called something new?

    ReplyDelete
  5. Oh and re PLoS One --- I am surprised they got away with some of this in a PLoS One paper. In a recent paper of mine there they made us take out any claims of "this is the biggest" or "this is the first" because they said PLoS One was about reporting what you did and unless we could document all previous studies we could not state which was biggest ...

    ReplyDelete
  6. To answer your question, the Syst Biol paper was invited, based on a conference "debate", and it included a list of questions that we were asked to answer.

    Others have written papers specifically about barcoding in protists, algae, and fungi. Bacteria are different in the way DNA is used, as you noted. So, in that sense, it's not DNA barcoding and is not directly relevant to the point that it must be mentioned constantly. Do you mention the human genome in every environmental genomics paper?

    ReplyDelete
  7. Sure now some people talk about barcoding fungi and microbes. But really, how is this different from rRNA surveys? And why does it need to be called something new?

    Really? Well, taxonomy in animals, plants, fungi, etc., has a few hundred years behind it based for the most part on morphology. The rRNA work often uses *only* DNA data. DNA barcodes in eukaryotes are, by definition (according to GenBank as well), standardized, linked to a voucher, etc. People have used DNA in taxonomy in a different way in bacteria, sure. And people have used DNA to supplement eukaryote taxonomy, obviously. But a standardized system across entire kingdoms? That's new for eukaryotes.

    ReplyDelete
  8. #1 - I was referring to microbes generally. There was much work on rRNA environmental surveys of microbial eukaryotes before the barcoding stuff came along. So to say all the rRNA stuff was about bacteria is wrong. There was plenty on euks and archaea.

    #2 - Woese used rRNA sequencing of cultured isolates to study phylogeny of all organisms, not just bacteria. And he started doing it in the 1970s. And after a few years this was standardized. And there are 1000s of papers using rRNA sequencing or other genes in standardized ways to do taxonomy of all sorts of organisms. So again, how is what became known as "barcoding" different than this?

    #3 - Blaxter does generally a great job of linking barcoding and other studies using DNA to do taxonomy (e.g., http://filogeografia.dna.ac/PDFs/barcode/Blaxter_04_DNAtaxonomy.pdf). Many others however do not do a good job. I stand by my critiques of many in the field. I think they purposefully or ignorantly do not refer to other similar work with the effect of making their work seem more unique. It is unnecessary. And annoying.

    ReplyDelete
  9. #1 - I was referring to microbes generally. There was much work on rRNA environmental surveys of microbial eukaryotes before the barcoding stuff came along. So to say all the rRNA stuff was about bacteria is wrong. There was plenty on euks and archaea.

    Environmental surveys are not barcoding, so the point is somewhat moot. Barcoding is, by definition, a standardized identification system that uses DNA sequences to identify species. DNA taxonomy is not the same thing. However, more recently "barcodes" have been used to identify clusters that can be considered putative species and considered on that basis in ecological or other work. But they don't get called "barcodes" in GenBank without meeting the strict criteria.

    #2 - Woese used rRNA sequencing of cultured isolates to study phylogeny of all organisms, not just bacteria. And he started doing it in the 1970s. And after a few years this was standardized. And there are 1000s of papers using rRNA sequencing or other genes in standardized ways to do taxonomy of all sorts of organisms. So again, how is what became known as "barcoding" different than this?

    1) Barcoding is not phylogeny. 2) Barcoding is comprehensive (i.e., there are efforts to cover entire taxa, not a broad sampling). 3) Barcoding is not used to delimit new species by itself, it is calibrated against existing taxonomy and putative species are subject to formal description as well. 4) Animals (the initial focus of barcoding, now including plants and fungi) have never had a standardized DNA based system for identification.

    #3 - Blaxter does generally a great job of linking barcoding and other studies using DNA to do taxonomy (e.g., http://filogeografia.dna.ac/PDFs/barcode/Blaxter_04_DNAtaxonomy.pdf). Many others however do not do a good job. I stand by my critiques of many in the field. I think they purposefully or ignorantly do not refer to other similar work with the effect of making their work seem more unique. It is unnecessary. And annoying.

    Exactly. Because what Blaxter does with nematodes is very similar to the work on microbes, in that there is no existing morphological taxonomy and very few morphological characters to base a taxonomic system on. It's DNA taxonomy and he even uses a similarity threshold approach similar to microbial studies. So it's very appropriate for him to draw a connection. Barcoding in the main sense is a means of identifying existing species and pointing out possible new ones that can be formally described by taxonmists. It's controversial enough, I can't imagine how it would go over if people just scooped up fishes and said "there are 50 MOTUs in this net, but nevermind formal names" and left it at that, rather than what FISH-BOL is actually doing.

    ReplyDelete
  10. Still not getting it Ryan.

    How is barcoding different from things like taking a collection of strains of microbes for which their is some taxonomy (based on chemistry, morphology, pathogenicity and other phenotypes) and sequencing rRNA genes from them and trying to define cutoffs to use to delineate what is species 1 vs. species 2 and then using the sequence instead of the phenotypes?

    This approach has been done with microbes for many many years. Sure morphological and other taxonomy works better in many cases in plants and animals. But molecular approaches have routinely been used to supplement phenotypic data as part of the formal description process for new taxa. This is even codified in the IJSEM and other organizations.

    See for example http://www.ncbi.nlm.nih.gov/pubmed/10551170 -- just something I pulled out almost randomly. Or look at hundreds of papers by Stackebrandt like http://www.ncbi.nlm.nih.gov/pubmed/1693659.

    Since the 80s people have been trying to develop calibrated, standardized methods to integrate molecular data and phenotypic data in the identification and classification of species of microbes.

    The fact that people are unfamiliar with the work in microbes and think that all work in microbes is sequence based is part of the problem here. Barcoding does not seem to me to be significantly distinct from prior work on microbes.

    ReplyDelete
  11. Quick! Alert Williams and Ebach! They will no doubt wish to include this in the next edition of Foundations of Systematics and Biogeography.

    Meanwhile, I'm going to check to see if I have any coupons for Maalox.

    ReplyDelete
  12. The fact that people are unfamiliar with the work in microbes and think that all work in microbes is sequence based is part of the problem here. Barcoding does not seem to me to be significantly distinct from prior work on microbes.

    The fact that people are not familiar with the background and current research in barcoding is also part of the problem. :-)

    ReplyDelete