Story: The Human Genome Project began in 1990 with federal funding and took 10 years to map out 1 full human DNA sequence. Toward the latter part of that methodical exercise, Craig Ventnor formed Celera Genomics (CRA) to do the same thing using a 'shot-gun' approach, that instead of progressively, methodically, and accurately reading and transcribing a DNA strand, chopped it up into tiny bits, transcribed as fast as possible those tiny bits, then put them back together in the right order using complex pattern matching algorithms based on Single Nucleotide Pairs [SNP's].
Sounds impossible, and when Ventnor proposed it to the original Genome Project, it was rejected, which stimulated him to form Celera, and ultimately to beat the Genome Project and prove the validity of his approach, which has now become standard. Human Genome Sciences (HGSI) did the same thing, and both were high flyers based on the promise of the commercialization of their results.
Simultaneously in the pharmaceutical industry, drug design went away from 'rational design' to 'combinatorix'. Rational drug design uses the insights of chemists to pick target compounds that may have efficacy against a particular disease mechanism. It's very hard work, requires smart people with lots of experience and insight, and can not be standardized easily. It is also largely responsible for everything we have in new and novel drugs, which make the world an unbelievably better place than it was only 50-100 years ago.
Combinatorix came up in the early 1990's with the advent of automated chemical manufacturing and diagnostic tools (e.g. Affymetrix), and used a somewhat different approach. Given a particular target compound, 1000's to millions of possible variants would be screened for binding or other interaction with the target, and then only those showing the expected screening would be pursued. The thinking behind combinatorix was rational and complex:
1. Many drug variations require only small side chain modifications for drastically increased efficacy, so performing those modifications automatically makes sense.
2. Doing anything systematically in drug development holds the promise of vast rewards, unlike the current lottery system where years of rational labor may very suddenly lead to nothing, or to a blockbuster. Rationalizing and systematizing the research processes could lead to more stable business.
3. The necessary computing power and other tools to pursue this idea had not been previously available at the right price, so the idea had to be pursued.
4. Rational design gets locked into certain ways of thinking, and not applying those rules may jump out of that limit-cycle of thinking and find new interactions.
The drug industry pipeline took a major nosedive during the 1990's, from which it has only recently begun to recover, and much of it is blamed on the failure of the combinatorix idea to live up to the hopes. The reasons are complex - a simple initial screen doesn't tell you much about the whole disease chain efficacy, the number of possible compounds is enormously bigger than the number that can be screened (for the same reason, with only 26 letters, we have not yet run out of words), etc. - and not vital for the current discussion.
So, combinatorix as a stand alone technique has failed, and it is now largely used, if at all, as just one adjunct to the rational drug design approach. But given the combinatorix excitement among scientists, one can understand how that same thinking would translate to the fruits of the Human Genome Project. So the question for Celera and Human Genome Sciences was always: How are you going to make any money? The answer was multifold, involving selling access to the DNA databases, and assisting in targeting drug design.
That last part is the big deal, and the whole promise of DNA research -- better drugs targeted to patients with particular genetic profiles. The general way companies trying to do this talk about it is they take the DNA profiles of a particular disease-expressing cohort of patients and compare them to a disease free cohort, looking for genetic markers that distinguish them. Those markers then become the drug design targets. You can see how that's a combinatorix-meets-DNA approach. There are some variations in how cohorts are selected, etc., but the approach is always 'do some statistical correlation studies of DNA samples to find possible drug targets.'
The big point: It sounds fantastic, and has huge promise, but has been tried since at least 2000 and nothing really has come of it. That's the nature of research -- mostly failure and dead ends.
Company: Given the sordid history of these efforts, coupled with the large failure rate in drug design, we were hoping that Perlegen, like Affymetrix (of which they are a spin-off), would be something like a tool company that sells to all the other players, but rather they seem to be pursuing their own drug designs, and assisting in clinical trial efficacy enhancement (just like the other players).
I fail to see what they bring to the table that is novel, although there are so many subtleties involved in successful research that there is almost always room for another team. But the most likely thing is that they will fail, so the question is whether they have jumped over those odds by already proving viability.
According to their initial prospectus the answer is No. "We have incurred $153.1 million in cumulative net losses since our inception in 2001, and we expect losses to continue for the foreseeable future." They are just another biotech gamble, with bigger risks than average, but also bigger - if however unlikely - potential rewards. If you are into biotech gambling, they are as good a dice roll as any. That is, there is absolutely nothing predictable nor assured about their financial returns. Sort of a Dog and sort of just Blah.
Stock: They have filed to list under PERL. For comparison, Human Genome Sciences (HGSI) and Celera (CRA) have been flat for 3 years, the current market for biotech IPO's has been brutal, and Symyx (SMMX) - a similar Affymetrix spin off focusing on using combinatorix to develop useful inorganic compounds - has not had a particularly interesting stock pattern for the last couple of years. The Genome Project excitement of 2000 is no longer, and based on other biotechs coming public recently, we expect PERL to be boring. Pass.
[Disclosure: currently long SMMX]