Monday, June 14, 2010

The Fired Coach Syndrome

The Fired Coach Gets Hired Again
Well, sports fans, you probably are all familiar with the Fired Coach Syndrome. That's the regular pattern whereby, when a coach is fired because his team doesn't win--even if it's because they haven't got good enough players--he is immediately hired by some other team and treated as their savior. We need not name names, because as one coach said, all coaches have either been fired, or will be fired (with the single exception of our own Joe Paterno, shown at left!).

What has this got to do with genetics, evolution, or any other MT theme? The NY Times has a story--a front page story--on the failure of genome-wide association studies (GWAS) to lead to many disease 'cures'. Big news!

Or is it?

This story is by an op-ed writer who has built his career over many years by serving as a notorious mouthpiece for the genomics 'industry'. Now, without a whiff of contrition, he's telling a new tale, and again making a Big Story of it. But, as before, he does it uncritically, too.

Genomewide Association Studies
It's true that countless millions have been spent on GWAS with little of the long-promised 'cures' to show for it. A few of us have been long-term critics of this approach, for the decade in question, and for legitimate reasons. Complex traits are here today because whatever their genetic basis, it passed the sieve of evolution. For human geneticists, the main interest has of course been the genetic basis of disease, for obvious societal and funding reasons.

But for many decades it has been clear that while genes contribute substantially to their variation, these traits (normal or otherwise) are basically not due mainly to single genetic effects. The evolutionary assumptions that are implicit in complex disease mapping studies never, except for wishful thinking, predicted other than what we have seen. And another fact that the NYT story and a recent journal article or two report as curious--that we can predict risk better with a patient's family history than with his/her genotype data--is entirely expected, based on reasons that everyone properly trained in genetics should have known (for approximately the last 100 years based on the genetics of 'polygenic' traits).

The Evolution of Complexity
Evolution generates redundancies, removes nasty variants, and generates a distribution of mutation effects and allele (genetic variant) frequencies that will not as a rule lead to common, chronic, late-onset, or complex traits being caused by just a gene or two (and here we're not eve considering the wild card, the environmental effects that are usually substantially more important than the genetic ones). But that doesn't mean there would be no identifiable contributing genes: for most traits some alleles, in some genes, in some populations, will have marked effects. Unlike a complete 'bust', that, too, is what we find and was predictable.

From countless GWAS and other approaches to enumerate the genetic cause of complex traits, we know of a large number of contributing genes, say around 1000. But only a surprising few of these do so replicably or with high probability and nontrivial frequency. What we see today is what everyone should have known to expect and a few of us were saying this in papers at the beginning of the decade in question.

Selling the Brooklyn Bridge--Again
So if there's no news in this news, why does the former hawker for the vested genomic interests now have the creds to write as if insightfully about the current state of affairs? It's by consulting the very people who, mainly knowingly, lobbied and hyped everyone into the GWAS era. Why should he be getting assessments of the situation from those whose mouthpiece he was for an approach that didn't work, and for predicted reasons? The same vested interests are now advocating their new grand theories, such as that finally we'll solve the problem in terms of (take your pick) copy number variation, epigenetics, whole genome sequences, large biobanks, many very rare alleles with strong effect, combinations of common alleles with modest, effects, etc.

Why should those who sold us the old Brooklyn Bridge be listened to when they advocate these new ideas which, not incidentally, require larger, longer, more costly studies (for their own labs?). They are not giving us reasons, just hand-waving invocation of current fads or Plan B's. Naturally, they like the idea of locking in funding for the rest of their careers. But would people have any interest at all in buying this new Brooklyn Bridge?

The FCS
The answer is the Fired Coach Syndrome. Somehow, we tend to believe that these people with known rap sheets are still the experts. Their past statements and advocacy are not taken to be the lobbying that they actually were. Now, they're being listened to as the coaches of the new Team GenomeSequence.

But there are deeper and more serious issues. Unfortunately, they are reflected in the latest Big Story now being marketed in this Times article and elsewhere, which continues to suffer from uncritical hyperbole and oversimplification.

We Already Knew All This
The fact is that after a few GWAS (and other kinds of studies of variation underlying single-gene traits), many years ago, we confirmed the theory we had about evolution and complex genetics (which goes back nearly 100 years). We should have declared our knowledge firm, and thought of better ways to understand complex traits. But too many vested interests, without better ideas, demanded the big GWAS funding. In that sense GWAS and its fellow-travelers have generally been a very expensive bust.

But here we have to again criticize the attempt to make this a new Big Story. Because we have learned a lot, even if at great cost, about genetic control. Huge and useful data bases have been created. Technology has been developed. DNA sequencing is now about as cheap as your HDTV set. Many corporations have flourished--money for their stockholders, and jobs for employees. Much has been learned about genomes of many species, and their variation. Despite private greed, much or most of these data are freely available to anyone. Genetics has flourished perhaps like no science ever before.

Maybe the funds should have been spent in other ways, but they have led to much new knowledge. This doesn't mean that 1000 new promising targets for Pharma have been revealed, as Francis Collins and Eric Lander enthusiastically claim: instead, most of those genes are trivial contributors. Still, these investigators have sponsored or done technically excellent work, contributed to the huge public data bases, new understanding of genomic functions of many kinds. They've done this while, largely unrecognized even by themselves, showing that classical theory was right--and that is a substantial baby in the bathwater.

There are important questions about complex causation, which science can address, especially if we could decide to confront complexity on its own terms rather than promising to reduce it to a manageable number of enumerable causes as has been done to date and is essentially still being done.

However, the reportage errors continue: Despite the waste and the catering to vested political interests, the Times article's main claim to Big Story, the fact that we haven't developed many 'cures' of complex disease, is irrelevant to whether GWAS was a scientific success.

Disease is complex, organisms evolved to resist tinkering from the outside--especially tinkering with our genomes--and attacking diseases genetically is a difficult engineering feat. Whether, when, or how that will be done successfully is not yet clear, but nobody has any right to expect it to be rapid. It's a bum rap at GWAS to say they failed because the gene engineers haven't yet figured out how to use the results to make cures. The real rap, and what should discredit much of the hype machine, is that they have promised 'cures' (in fact, and to be fair, some of the main advocates--though not Francis Collins--have at least been slightly circumspect, warning this is all for our grandchildren or beyond--even if such caveats were declared in passing and did not temper their lobbying for the GWAS and related resources).

Bigger, Longer, Pricier More-of-the-same
However, the Failed Coach Syndrome would suggest that we should be very skeptical about the same writers and scientists who are now deftly expostulating their latest grand theories, paradigm shifts, and new strategies. Because they're largely rationales for more-of-the-same, except at larger scale, longer time frames, and greater cost. What we should be doing is understanding where the baby is, and not just fostering existing self-interest by saying there was no baby in the bath so let's run a lot more bathwater and maybe we'll find triplets in there someplace.

One thing that some are advocating is to search genomewide sequence data from patients, to find clearly harmful mutations in functional genome regions that are known to be related to the physiology in question--such as rare 'knockout' mutations in those genes which might be inferred to be causal and then tested experimentally. Whether this justifies the continued use of mega-scale genomic approaches is debatable. Though one can predict there will not be a huge bonanza of findings with much therapeutic value, there certainly will be some, and at least to some extent this genome-screening approach is very different from the statistical-evolutionary basis underlying GWAS. The reason is too complex for this post, but the gist is that such studies will finally rest on biology rather than baloney, even if it's already being over-sold in the usual lobbying way.

Today's PT Barnums
As to Fired Coaches, it must be said that high level expertise and capability is not to be found everywhere, and our genetic PT Barnums got there largely because of both management and scientific skill. Once the research community has its hands on funds, and labs have been built, it's politically difficult if not impossible to shift from this entrenchment to new investigators or unrelated approaches. Science is part of society and works by evolution, which includes careerism. Scientific revolutions can't be ordered up by the media or by funding institutions with their turf-protecting bureaucrats: they just happen when and where they happen. So in that sense we must in practice rely on the same cast of characters, though whether we should try to move gradually away to fresh approaches is a legitimate question.

In science, we should at least be aware of what's afoot because otherwise resources continue to be diverted for less rather than more effective use. But can we learn to temper our claims--or at least penalize those who don't? Or can the system demand accountability for results if we promise them? Or can we establish full-stop criteria for large projects once it's clear that they aren't really bearing fruit? Probably not. Because the real name of the game is getting funding. To a substantial extent, scientific facts themselves are secondary. That's the anthropological truth. Or perhaps, as Marshal McLuhan said decades ago, in modern science the medium really is the message. And now we are clearly going to hire the same coaches to guide us into the future!

Maybe we have no choice. But we should be aware that's what we're doing. And if you believe their new pronouncements or their media megaphones, then we have a bridge that we'd like to sell you. And, by the way, Joe Paterno has stayed in the national rankings and just signed a new contract....at age 83.

2 comments:

Joe Terwilliger said...

Ken, I highly suggest you do as I do, sit back on the couch with a big beer, a smug grin, and bask in knowing that despite their pathological inability to admit they were wrong and we were right, we all know that Lander, Collins and their ilk realize it, and that it must be killing them ;)

Ken Weiss said...

Well, the issues are subtle as we tried to show, but subtlety doesn't snow up in the news, and we still face enormous biobank/whole-genome efforts that are being justified as needed because of the success (not admitted failure) of GWAS. So nothing's 'killing' anyone because of what has been discovered.

If it really were discovery, rather than expensive demonstration of what we knew to expect, then I'd think one should listen to the same people as they outline what they want for the future.

And again, the real science--what we learn about causation--needs to be judged on its own rather than in terms of the lack of quick cures. As we said, Collins and others should not have promised miracles, the reporters should not have repeated them uncritically.

None of this takes away from the basic low yield of high-cost association strategies, however, and that shows the importance of a knoledgable evolutionary perspective.