Wednesday, November 12, 2008

Underselling Genomics Award #1: David Whitworth for "Genomes and Knowledge: A questionable relationship"

I do not normally write too much here about non Open Access publications but this one is so good I had to. Everyone with access to Trends in Microbiology should check out Julian Parkhill's rebuttal to an article written by David Whitworth in the same issue. Whitworth's article is "Genomes and knowledge – a questionable relationship? " and it is in my opinion, filled with some unsuppoerted and over the top statements. In essence, he is arguing that we should stop genome sequencing because there are a bunch of genomes out there already and after all, all that matters is work on model organisms so if you have enough genomes related to your model organism you should move on. Alas I do not have time to detail them here. But fortunately, Parkhill does a great job of responding in his article Time to remove the model organism blinkers. The end of his article reflects how I feel too:
In the end, when faced with the astonishing diversity of microorganisms, if all we manage to do is to describe a few random organisms in painstaking detail, then we will have failed to understand microbiology. To suggest we curtail the remit of microbial genomics is bad enough; to suggest it now, when we are on the brink of finally being able to truly study genomic diversity, is absurd.
So sure, sometimes we in genomics oversell the benefits of genome sequence data (and in fact, I give out a little award here for those people). But Whitworth is at the other end of the spectrum, wearing, as Parkhill states "blinkers" to the benefits of genome sequence data. As a reflection of how much I disagree with most of Whitworth's implications, I am giving him my first "Underselling Genomics Award".


  1. A well-deserved award.

    Back when HGP was still not finished and there was all this hype about the Holy Grail and Blueprint Of Life and such, I liked to pour cold water on the hypsters about it. But one thing I always emphasized was that HGP is necessary in order to make the sequencing techniques good, cheap and fast so we can start doing what really matters: sequencing genomes of every critter we can get our hands on and thus starting an era of a truly comparative biology, one that does not rely on just a dozen model organisms.

  2. BTW, can you send me the second paper by e-mail (the one by Parkhill), please?

  3. I disagree with Whitworth's article even from the perspective of studying model organisms. Automated methods for predicting gene function, such as "gene neighbor" or conserved context, are expected to perform better with large numbers of distantly related genomes. Identifying covarying residues to predict protein interactions, or subfamily specific residues to illuminate gene function, is expected to work better with very large alignments. Identifying regulatory sites with phylogenetic footprinting works better with large numbers of closely related genomes. Furthermore, because of horizontal gene transfer, there is no a priori way of knowing which other bacteria will turn out to encode homologs of your gene of interest, which may help highlight its function. In summary, the idea that 100 genomes suffices to interpret the genome of Escherichia coli K12 seems almost as absurd as the idea that 100 genomes suffices to illustrate the diversity of bacteria.

  4. I obviously agree with you, JE, but it's hard in practice to argue against people who don't get the point of genomes, because the very people who understand genomes the best are those involved in their sequencing and analysis -- you have a connection to JGI, Parkhill is at Sanger, and I'm at JCVI. So it's easy for critics to sneer and just say "Well, of course *you* would defend genomics".

  5. other Jonathan --- with no other information you are correct --- however, I spend much of my time (and a lot of words on this blog) railing against people who oversell genomics --- and I will continue to do this --- so I think I have some credibility in saying I do not think genomics solves all the worlds problems --- but at the same time, Whitworth is basically suggesting that Wellcome, NIH, NSF, DOE, Moore, and tons of others have all been swindled and are wasting their money --- whenever you directly or indirectly criticize a huge collection of people like that it would be good to have a bit more evidence to back it up

  6. Right -- but still, I can't help thinking that the best rebuttal to Whitworth would have been from a non-genomicist whose research program was based on things learned from those "redundant" genomes beyond the first 100...

  7. yes, sure that would have been better

    but I guess I was trying to say this ... the program officers at NIH, NSF, Moore, DOE, Wellcome, etc, they are not genomicists and yet they fund this. You may not like program officers but can all of them be completely swindled?

  8. One of the things I do is teach metabolism to undergraduates in an introductory biochemistry course. I also write their textbook.

    In order to appreciate cell metabolism we have to understand how common the various pathways are and how they evolved. One of the spectacular results of genome sequencing is that it gives us a window on all the pathways in the cell.

    Comparative biochemistry is expanding as a result of the genome projects.

    Having multiple genomes tells us a great deal about which pathways are fundamental. Here's two things we learned from having the sequences of many genomes.

    1. Some bacteria are missing the enzymes for glycolysis but all of them have the enzymes for gluconeogenesis. This strongly suggests that gluconeogenesis came first and the glycolysis pathway evolved later.

    2. Most species of bacteria do not have a complete citric acid cycle. This allows us to reconstruct the evolution of this pathway and demonstrate how irreproducibly complex structures can evolve.