They're back. Investors - who a few short months ago gave little credit to companies launching new products - are increasingly willing to award substantial increases in valuation to companies reporting early clinical or even preclinical data, a sure sign that the biotech sector is at last on the upswing.

The most dramatic evidence of the new attitude was last week's reaction to the publication in Science of preclinical data on the ob gene, which Amgen Inc. has licensed from Rockefeller University for $20 million. Investors added $735 million in valuation to the stock by week's end - it was up as much as $1.26 billion - and the stock closed the week up $5.625 at $85.625.

That's $735 million for a protein that's not yet in the clinic, in a field that's not well understood, where there will be competition from multiple products, and where there will be multiple opportunities as the compound progresses through the clinic to trade in and out on milestones.

It doesn't detract from the major strides that have been made in understanding obesity in the past year to note that the role of the OB protein won't be as simple as, say, human growth hormone or Factor VIII, where replacement therapy for patients with a deficit was the obvious strategy.

Now that investors have had their fun, there's time to take stock of what is and isn't known about the role of the OB protein in obesity. The three Science articles looked at the effects of mouse OB protein and recombinant human OB protein in ob mice lacking both copies of the ob gene; in lean mice with either two copies of the normal ob gene (+/+) or one normal copy and one presumed mutant copy (+/?), in db mice with the diabetes mutation, and in mice with diet-induced obesity (DIO).

In a nutshell, the following is known:

- OB reduced weight in ob mice and normalized their insulin and glucose levels.

- OB didn't significantly reduce weight in older +/+ mice, mice that might be analogous to obese adults.

- At higher doses, OB reduced weight in younger +/? mice, mice that might be analogous to obesity in young children.

- OB didn't affect db mice.

- OB was ineffective at low doses in DIO mice, but worked at high doses.

The following isn't known:

- The receptor for the OB protein.

- The role of OB in humans.

As noted in the editorial in Science accompanying the three articles, results in the ob mice don't necessarily translate to obese humans, because thus far it appears that human obesity isn't as simple as a flaw in the ob gene.

Multi vs. monogenic

Based on what is currently known, human obesity appears to be a multifactorial and multigenic condition, whereas the mouse models have monogenic flaws. The ob mice lack the protein itself, while the