Irrational drug design (aka high-throughput screening) parallels other areas of data-driven science: it abandons the methodology of traditional hypothesis-driven science — which demands a focus on specific predictions — and pursues instead the weak and humble hypothesis that looking in a general area will find something. As I discussed here, genomics and synoptic sky surveys are classic examples.
High-throughput screening for drug discovery differs from these examples in that the areas studied are real only in an abstract sense, consisting of regions of molecular-structure-space that nature hasn’t filled. And indeed, making and testing millions of drug-like molecules often turns up interesting results, in both a practical and a scientific sense.
A few of the 1000s of new anti-malarial drug candidates.
Two papers in Nature (here and here) turned this method on malaria, finding thousands of candidate drugs, defined as molecules that inhibit the growth of malaria parasites by 80% or more at concentrations in the micromolar range. The hit rates were, respectively, ~1,100/310,000 and ~13,500/2,000,000.
Many of these candidates are unlike any now in use (different structures, different mechanisms of action) and crucially, these often evade mechanisms of resistance to existing drugs — a problem that threatens to restore malaria to its previous level of deadliness across much of Africa and South Asia.
Promise beyond malaria
In our household, small risks of malaria arise only when traveling to these regions. A footnote to the search for new malaria drugs, however, indicates the potential for finding treatments for scourges of the older populations of temperate zones: the disorders of protein hyperphosphorylation that underlie the leading dementias, including Alzheimer’s disease.
These result from imbalances in the activity of enzymes that add and remove phosphate groups to alter the activity of proteins. Kinases do the addition side of this, and they are diverse and specific. Many of the candidate drugs discovered in the studies reported in Nature are thought to work by inhibiting specific kinases in malaria organisms. Molecules with similar discrimination in modulating the activity of specific kinases in human cells could provide leverage in a treating a range of human diseases, including those that can steal the mind.
See also:
{ 3 comments… read them below or add one }
Exploring a new and complex solution space probably has to be done by blind luck. I suspect that a lot of science happens that way, even in more traditional disciplines. A lot of good ideas are rejected for no good reason, but simply because of an intuition that they are unlikely to work – an intuition that is likely true, but largely unrelated to the merits of the particular idea.
I’ve even speculated that the high rate of idea-rejection in science is adaptive – that although a mere trickle of good ideas are (more or less randomly) established to work, these are enough to keep science busy in mapping the new regions of solution space opened up by them.
This implies that science collectively, and scientists individually, must not be aware of why they’re rejecting the ideas. And we see this in several cases, including molecular manufacturing.
At the FNANO conference, one scientist after another told me that they would never have expected Rothemund’s origami technique to work – and they cited reason after reason that I think, even without the benefit of hindsight, could have been shown to be incorrect with a few minutes of creative thought.
So perhaps irrational drug design and data-driven science are merely ways of doing what science has been doing all along: stumbling forward blindly, adopting the few things that seem to work, and then filling in the gaps over time.
I wrote more about this here:
http://crnano.typepad.com/crnblog/2010/04/scientific-advances-require-blind-luck.html
Hey–rational drug design and hypothesis-driven science miss a lot. Screening is by no means “irrational” or “stumbling blindly”–a better word is UNBIASED. You develop a test for an important biological output, without many preconceived notions about what the compounds are supposed to look like and the target is supposed to be. You let the biology guide you. You are not a jerk about it thinking you can tell the biology what to do.
Many of the drugs on the market today were discovered by a combination of design and serendipity, and the idea that everything should be planned out in advance is ludicrous. The alternative to data-driven science is ego-driven science.
Warren — I agree. I’ve seen the term “irrational drug design” used, but always in a humorous way that pokes fun at the initial bias against the concept. I like both rational design and humble exploration guided by only the most minimal hypotheses: Both of them are effective, and interesting, too.
The evolutionary problem-solving intelligence of nature is a grand manifestation of highly effective but non-rational design — blind watchmaking, as Dawkins calls it.