Expect miracles, be disappointed
August 25, 2010 § Leave a comment
Ars Technica’s Nobel Intent blog has a long post from Robert Fortner today on the “declining expectations” around systems and synthetic biology. The expectation for systems biology, apparently, was “to produce a realistic cell entirely in silico” and “models that precisely predict biological events—like disease”, and the author is disappointed that these small goals haven’t been accomplished in the 7 years since Lee Hood founded the Institute for Systems Biology. Oooo-kay. And synthetic biology has “hit a ceiling” because for 4 years running the number of components used in synthetic networks hasn’t increased above 6.
I don’t disagree with Fortner that systems biology and synthetic biology were over-hyped in the first flush of enthusiasm about the possibilities of these new fields, and that the outside observer might therefore feel disappointment at what has been achieved. I don’t think that’s an uncommon pattern when new fields of science emerge, or new technologies emerge. I seem to remember similar things happening to nanotechnology, structure-based drug design, genomics and the internet, before expectations became more realistic and the technologies began looking exciting again. The phenomenon has even been given a name — the “hype cycle“— and Harvard Business Press has published at least one book about it. (And yes, the hype cycle has itself been accused of being hype; but that’s more about the analysis of trajectories than the basic phenomenon that early claims for a technology tend to be overblown.)
But what a shame to evaluate systems biology and synthetic biology this way. The claims quoted are from people who were trying to feel their own way to a vision of what could be learned, at a very early stage of thinking about the possibilities. Some of it seems odd in retrospect, if indeed it’s accurately quoted — to me, the goal of “models that precisely predict biological events” sounds unlikely to be met — what in biology is ever precise? (well, first let’s define precise). But that doesn’t mean that we won’t have models that help understand what’s happening in disease and why, and help predict whether drugs will be effective. Fortner seems to feel that the only goal that matters is a direct mapping of genotype to phenotype, and is particularly waspish about what he claims is a lack of progress on YeastNet.
Here’s the quote:
“Edward Marcotte, who leads the effort at the University of Texas, said in March that his team had updated YeastNet. He claimed to see “a nice increase in predictive power over v. 2,” however he has “yet to write it up or release it,” suggesting that, even with improvements, the results might not be so prepossessing. YeastNet’s publication arc has tailed down, beginning from the commanding heights of Science, descending in version two to the egalitarian plain of PLoS ONE and now apparently a file drawer at UT Austin. Similar difficulties beset the Alpha Project, which repeatedly scaled back its ambitions until it winked out of existence in 2008.”
So, Ed Marcotte must be fibbing about the increase in predictive power he sees, because Science didn’t want v.2? That’s a bit of an extrapolation.
Fortner says that “the outpouring of data has failed to coalesce into a solid theoretical foundation from which to build”. Well, on the one hand, yes — there is no ultimate theory of biology; and on the other hand, no — there are definitely places where collections of data are turning into knowledge. Here’s one that comes to mind, from Vamsi Mootha‘s lab: integration of genomics and proteomics data via machine learning led to a prediction that 1098 proteins are involved in the function of human mitochondria, most of which had not been thought to be associated with mitochondria before. One of these novel predictions led to the identification of the genetic defect in a familial form of mitochondrial deficiency, which seems like a good test of the validity of the predictions. I’d say this work looks like a fairly solid foundation (based on both theory and experiment) from which to build an understanding of the mitochondrion in health and disease.
I’m being a bit unfair to Fortner, whose thesis is not merely that nothing amazing has happened yet, but that the complexity of what is being attempted is going down. Possibly true. But I think this misses the point: the head-on modeling of whole cells or organisms is not at all the whole story of systems biology, and not even something that everyone in the field agrees is a good goal. The whole question of what kinds of models are helpful is an area of active discussion, and what’s useful will undoubtedly vary depending on the biological problem. Sometimes highly detailed models will be important; sometimes the best way to make progress will be to miss out a lot of the details, and aim instead to identify the main forces that drive the biological behavior.
Although I find the article rather disappointing in some ways, I should acknowledge that Fortner is absolutely right that modeling biology is complex and challenging, possibly more challenging than anyone expected in 2003. What I disagree with is the implication that there’s no progress. But then, I would, wouldn’t I.
For a different perspective on the same article, see this post over at the Finch and the Pea.
Update: Ed Marcotte has offered a response to the original article.