Is drug discovery moving too fast?
I'm a big fan of the singer Jack Johnson and for some reason I've recently had the song Inaudible Melodies in my head. The song references a story about martial artist Bruce Lee, who apparently moved so quickly that a director for one of his early films asked him to slow down. The reasoning was that he was moving faster than the individual film frames in the camera and that made his action look blurry, or out of focus. The song chorus goes like this:
Slow down everyone
You're moving too fast
Frames can't catch you
When you're moving like that
In this month's issue we have a couple of stories written by Randy Willis that bring into focus, if you will, why this song popped into my head.
The first story is a partnership between FEI and Swedish company Sidec Technologies whereby the two intend to commercialize tomography platforms based on FEI's transmission electron microscope and bring it to the drug discovery market (see page 16).
High-throughput screeners must have nightmares about the relatively slow speed of electron microscopy. In today's drug discovery world of increasing speed, scads of data and compound libraries of a size that boggle the mind, the focus is and continues to be not on slowing down, but speeding up and trying to find a better sieve.
But the folks at FEI and Sidec will have nothing of this. To their thinking, failures at the trial level are perhaps the result of too much Bruce Lee in the screening of compounds. Their tomography solution is potentially one method to bring things into better focus.
As Hans Johansson, Sidec CEO and president sees it: "Understanding drug targets and the disease mechanisms on a molecular level is extremely difficult, resulting in severe problems with translating bench research into human biology. Many drugs fail when they progress from preclinical studies to clinical trials. Imaging and comparing drug targets from different species using protein tomography provides insight into the molecular differences underlying this attrition."
In other words, HTS can only get you so far, or so close to the target. Perhaps, real understanding and better candidates can come from taking a breath and slowing down just a touch, in order to better understand compounds before they head their merry way down that NDA, IND and clinical trials path.
A second story that came to mind was the work recently completed by researchers at Johns Hopkins, Germany's University of Würzburg and the Institute of Bioinformatics (IOB) comparing protein-protein interactions (see page 12). Their work studied interactomes from human, yeast, work and fly datasets in an attempt to find overlaps and missing data sets as clues to metabolic pathways. One surprising finding of the research was the relative lack of overlap between the species, which flies in the face of the assumption that there would be clear parallels between drug impacts on mice or yeast and humans.
The researchers chalked this up to either incomplete data sets or, worse, faulty data sets, hastily compiled using automated techniques. Dr. Yuri Nikolsky, CEO of GeneGo, puts the problem rather succinctly when he says: "This is why manual curation from small-scale experiments with direction and mechanism is so important."
Is this slower than automated data mining from published literature? Without doubt. But it seems to me, that at some point in the process of taking compound to drug, there has to be a breather, a time to take more time and to develop more precision and understanding.
There are any number of reasons why there were only about 20 FDA new drug approvals last year and it seems plausible that at least one reason is the almost maniacal passion for speed in the process.
Bruce Lee found a way to slow down, just a little, so the frames in the camera could clearly capture his movements. Perhaps drug discovery operations could learn from his cue.