I'm a big fan of open access, and I'm rooting for PLoS Biology to succceed in their goal to publish papers that have the same quality and impact as Nature and Science. Today, though, I find myself both surprised and disappointed, as PLoS Biology shows that its hunger for publicity threatens to turn it into a vanity publisher.
You see, in an article appearing tomorrow morning (4 Sept 2007), PLoS Biology is publishing a paper by Craig Venter and colleagues about....(pause)... Craig Venter's genome! So when I get enough money to sequence my genome, will PLoS Biology publish that too? Is this the Donald Trump-izing of science?
But wait, you might object - maybe there's some science in the paper. Well, first you might want to recall that the original human genome paper published by Celera, in early 2001, was predominantly based on Venter's DNA, as he later admitted in a letter to Science. (Disclosure: I worked with Craig and was a co-author on the 2001 paper, and I fully stand behind it.) So most of the science was published over six years ago.
Still, I have the paper here and I've read it, and I can re-assure you that there isn't anything new, unless you are interested in how many differences there are between Craig Venter's genome and the reference human genome (the one in Genbank, which is very nearly complete). If anything, it's kind of a weirdly voyeuristic read, in which you an learn details of Craig's family history (English ancestry, three siblings, etc.), and you can see a full-page picture of his karyotype. Then there's a lengthy description of how they compared the genomes and found all the differences, but I confess I couldn't go on beyond page 10. In what I read, there wasn't much new science, certainly not of the caliber I've come to expect in PLoS Biology.
PLoS Biology will probably get plenty of attention for this paper. In politics there's a saying: any publicity is good publicity. In science, though, we think otherwise. PLoS: why'd you have to do this? I guess I shouldn't have expected so much of you.
I too had an opportunity to read the manuscript prior to release and I had serious questions about the judgment of the editors of PLoS Biology and the referees who approved its publication. Like Steven Salzberg, I find little of scientific value in the work. And while one can raise a whole host of issues related to the science - or lack thereof - as well as to the motivation for doing this work, I have another, more serious concern.
ReplyDeleteAny federally funded study requires protections be put in place to safeguard patient confidentiality and so that private genetic information cannot be linked back to them - in large part because this could have a negative impact on them in many different ways. All institutional review boards (IRBs) require the use of anonymized samples for this reason and most go so far as to require that even such seemingly benign information as zip (postal) codes are stripped from sample records prior to analysis and certainly prior to publication.
Here, we know not only the identity of the person being analyzed but we know he as a son and three siblings. Now simple genetics tells us that these first-degree relatives share half of Craig's genome sequence and their identities are little more than a Google search away.
Certainly, Craig Venter had no objections to having his sequence published, but no mention is made in the manuscript about how the ethical issues associated with this work were handled. Were his relatives given the opportunity to consent? Were they informed? Was their protection considered? There is nothing in the manuscript to suggest that these issues were addressed.
There is a brief mention that there was an annual Ethical Review, but one has to question the judgment - and independence - of the group that was given this task.
While human subjects protection rules may not apply to privately-funded vanity research, one would expect the publishers - and the referees - to have held this work to a higher standard and to consider their responsibility to protect human subjects. Sadly, it seems they have not.
Personally, I think both of you have let your dislike of Craig cloud your judgement here immensely.
ReplyDeleteFor example, regarding the informing of relatives - why should one be required to do that? There is nothing currently that restricts anyone from announcing they have Huntington's disease or are carriers of CF or have HNPCC or any other genetic problem. All this is is more information and it is his to use as he wishes. I have discussed this issue with George Church and some of the people who have signed up for his personal genomics activity and this is the standard response. Yes, it would be good to discuss this with relatives. But do they need full government backed informed consent. I do not think so.
Regarding whether this is a PLoS Biology worthy paper - I think the data alone makes this PLoS Biology worthy (note I had nothing to do with reviewing or accepting this paper and did not know anything about it until last week and I have not read much of the paper so cannot comment on the science in detail). Steven this is the same complaint that has been raised about many previous first genome papers - why pick on PLoS Biology over this one? The fact is, I am virtually certain this paper in some form would have been accepted at any journal. I think whether this was Craig's genome or Watson's or Ester Dyson's or whomever's it would have been treated similarly. In fact, I would guess because of Craig's reputation they probably gave this paper MORE scrutiny than someone else's genome would get, not less, but again I had nothing to do with the review.
Jonathan, I respectfully disagree. I don't think the data alone makes this worthy of PLoS Biology - how can you make this argument? I've been involved in sequencing dozens of genomes now, and everyone knows that the 2nd version of a genome (a new strain of a bacterium, for example) usually doesn't get published in a top-tier journal. The reason is simply that most of the major scientific findings appear with the publication of the first genome for a given species.
ReplyDeleteIn your post you say this complaint has been made about many previous "first" genomes - that's true enough. There are some genome "firsts" that were rather poorly done, with analysis that wasn't very insightful or even interesting. But that is a non-sequitur here: this isn't the first human or even the 2nd human genome, since two different human genomes were published in 2001.
The science in this paper, such as it is, is little more than a description of how Craig Venter's genome differs from the reference strain. This is a genomics version of publishing your medical charts - fascinating to the patient, but of little or no interest to anyone else. And I've yet to find - or to hear from anyone else who's found - anything of particular scientific note in this paper.
As for the scrutiny that PLoS gave the paper, neither of us know what they really did, but I believe they wanted to publish it because they knew it would get publicity. I note that PLoS issued a press release specifically about this paper - something they certainly don't do for all publications. And we both know that it's easy enough to pick reviewers who will give a paper like this a positive review.
I also disagree with your contention that this is about Craig personally - I would make exactly the same argument if it were Jim Watson's genome (which has also just been sequenced) or anyone else's. In fact, if a top journal publishes a paper on Watson's genome, and if there isn't anything scientifically new described there either, I pledge to blog about that too - count on it.
I disagree with the notion that interesting scientific findings "appear with the publication of the first genome for a given species." It is the comparisons among closely related genomes (e.g., E. coli O157:H7 vs. K12) that have led to some of the most interesting scientific findings. And the same is true for eukaryotes in my mind - the comparisons of Drosophila and Caenorhabditis genomes for example.
ReplyDeleteBut in this case there is an additional added importance to this type of data. The previous genomes were composites. That is useful for laying out the general features of the human genome but it is not that useful for population and comparative studies. It is funny in a way the human genome studies have taken the reverse approach of microbial studies. They started with the "composite" genome approach which is akin to metagenomics. But they found, just as people have found with metagenomics, that is is very useful to have a template of an actual genome to compare to and not a composite. And so this genome data is going to have uses much like reference genomes have in metagenomics. Therefore, I think it is not accurate to say that there have been two human genomes published before and to discount this one. This data is fundamentally different. And it has value from the population genetic and comparative point of view that the previous composites did not.
As for the press release, I have no clue how they decide to issue a press release or not. But come on, you have to accept that it is kind of obvious that this would get a lot of press coverage so it would seem inane to not issue a press release.
And fine, I accept that you might make the same criticisms of other people's genomes if they came out in papers. But I still disagree with your assessment of the interest and importance of this DATA in and of itself.
Jonathan, I think you're still missing my point(s). First, you write that "the previous genomes were composites." Well, not really. As Craig himself has stated, the Celera human genome sequence was dominated by his DNA. There were 4 other donors, but something like 80% (if I recall) of the DNA was his. So it wasn't really a composite - and it was the same genome described in this paper. It is not "fundamentally different" as you write. And the public group's genome, although supposedly based on DNA from a dozen people, also came primarily from one person - an anonymous man from Buffalo, NY (ask Pieter de Jong).
ReplyDeleteBut wait: Venter's genome is a composite - as were the other human genomes - because the two copies of each chromosome are basically collapsed into one. Some of the press coverage (e.g., the Washington Post) makes it sound like they separated the parental genotypes in the Venter genome, which would be interesting if true. It isn't. I've read the paper, and the best they could do was create haplotype blocks spanning 200 Kb for about half the genome. [Note to non-geneticists: these are blocks where all the sequence comes from just one of the two chromosome copies.] Beyond that, they don't can't link these blocks together, so the chromosomes are indeed composites.
And recall that most of the public effort's human genome was sequenced in BACs, which are 150 Kb regions from a single chromosome. So roughly the same amount of haplotype contiguity has been available in that genome since 2001.
And let me revise my comment that the most interesting scientific findings "appear with the publication of the first genome for a given species." Of course that isn't always so - I didn't mean to imply that - so I agree with you here. I agree that it's possible that a 2nd genome sequence (or a 3rd, or a 10th) could produce exciting new results. But what about *this* paper? I've read it, and I stand by my comments: this paper offers little in the way of exciting new science.
Jonathan, I too respectfully disagree with your assessment. As someone who works with human samples all the time, I know the responsibilities we all have to safeguard patient confidentiality and the need for anonymity both for the sake of the individual and his/her family members. Many journals require statements regarding protection of human subjects and here, more than in most studies, this seems to be an important issue. In fact, the era of personal whole-genome sequencing opens up a new world of potential ethical considerations with regard to data release and publication. The failure of this to be discussed in the paper is an important oversight.
ReplyDeleteI will admit that you may be correct in noting that my previous interactions with Craig may have clouded my view of this. Most studies require independent review of experimental protocols involving human subjects. It should not surprise you, therefore, that I have questions about this aspect of the work. And it is for this reason that I am disappointed in PLoS Biology and the referees for not making a point of requiring more detail in the manuscript.
I would deem this publication "new" science worthy if PLoS Biology were to publish the results of a broader study that demonstrated the genomes of a segment of the geneology of an anonymous subject (Venter's family should tread cautiously). In the long term, yes, this sequenced genome will be one of millions, but at this point in history, if using flashy names to catch the attention of the public, investors and researchers into supporting and understanding further genetic research, let it be. Sadly, the name Venter may give it more of a chance of getting a 30second blurb in the media rather than the actual scientific value. They might as well use a more familiar/celebrity media name to secure public interest. More "selling out" in this way just might increase funding so that research grants would outweigh other government spending. Also, the ethical/ confidentiality concerns that are present in this attention- craving paper may set off alerts of the need for an improved system designed for this research. From this perspective, will the ends justify the means?
ReplyDelete