1. Hap on August 11, 2009 11:54 AM writes...

I'm not seeing anything about the 127,541(?) compounds that didn't come up as hits. I don't see a peep, a reference to a database somewhere, anything. Did I miss something?

2. John Timmer on August 11, 2009 12:13 PM writes...

You're thinking of this in terms of drug development, but i suspect the authors just included that as a pitch - it's mostly about the biology. If you look at the discussion, there's only one paragraph that mentions drug potential; the rest is basic biology

There's a reason for that - the exact details of Hh signaling downstream of the cell surface are really not well understood. There's complete degradation of the Gli proteins, partial cleavage that makes them into repressors, and some form of activation that converts them into activators. We don't have a clue what the chemistry of that last one is. Add to that that the Gli proteins get shuffled around the cillia for all this processing, and interact with a complex mix of kinases, ubiquitinases, sumoylation enzymes, and lots of scaffolding proteins.

So, basically, anything that can help us block a single step in the pathway and let us do some sort of epistasis analysis is a very valuable tool in the short term. Maybe someone could eventually make it into a drug, but that doesn't seem to be the focus of this paper.

3. anon on August 11, 2009 12:45 PM writes...

#2 - I think you're partially missing Derek's point. Even as a so-called "valuable tool" you're willing to believe these guys came up with could be dirty as all get-out against many, many different molecular targets. Without understanding the molecular target--and your compound's selectivity for said target--any of the biology dissected from that would be practically meaningless. The phenotypes the authors observe are just outcomes. Lots of combinations of different mechanisms can lead to similar outcomes.

To use an equivalent in the kinase arena, this could be like using staurosporine to inhibit your kinase of interest, and then concluding that all resultant biological effects are solely due to inhibition of your kinase of interest. Completely ignoring, of course, the fact that staurosporine inhibits almost every kinase known to humankind.

Incidentally, this post has some bearing on some of Derek's other recent posts about the academic/industrial divide and the public perception of how much industry contributes to the whole process. Academic groups coming up with tool compounds--like this paper--almost invariably fail to fully evaluate the complexity of the problem of target selectivity. Whether this is through lack of resources or simple ignorance, this level of sophistication is critical to the industrial drug discovery enterprise. Many of the biologists I work with spend their fair share of time debunking all the crapola published by academic labs because of this lack of thoroughness.

4. Chris on August 12, 2009 2:18 AM writes...

There is a common theme to the four "actives" identified in this paper. They are all commercially available compounds with a CAS registry number and (almost) no literature references. In each case there are commercially available analogs at high similarity again with CAS registry numbers and again no references. I frequently see this pattern in "actives" and it makes me deeply suspicious. What do you think is the probability that a vendor would make a totally novel series just to hit in my new screen? If I were suspicious I might think that the origin of each series was a compound with a flaw that hit enough screens to warrant preparing a flawed analog series. I particularly do not like HPI-4 with a push pull polarized double bond crying out "I am a Michael acceptor please interact with me".

5. anon on August 14, 2009 2:25 AM writes...

#4. How uncommon are the properties you describe for compounds in big commercial library's from chem div etc.