30 September 2013


Dan and I are were at the CHI Discovery on Target meeting last week.  It is highly focused on target validation and early stage hit generation.  This is NOT a chemistry conference, although there were plenty of chemists and chemistry talks; the target audience is biologists.  As such, it was a great arena to be teaching about fragments and educating a whole different phyla of FBDD consumers.  It was also nice to meet people and have them say, "Oh, I love the blog."  Of course, I just say, try commenting, that's manna to bloggers.  The nice thing is people generally understand FBHG, unfortunately I think they generally misunderstand it.  Why?

I think part of the problem is that the Most Impactful Papers (MIPs) in the field are also the Most Destructive Papers (MDP) in the field.  So, what are the MIP for this field?  For me, the criteria are pretty straight forward: one or two papers that are seminal to understanding the field.  As you may already be guessing, my list of MIP intersects my MDP.  

Most Impactful Papers
The Rosebowl of Fragment Papers: SAR by NMR.  This is the paper that showed that NMR was not bound by doing structures, but was a viable screening paradigm.  It started the whole "Fragment" thing. 
  • The Rationale Behind it All: The Leach and Hann Molecular Complexity paper.  If I had one paper to give to someone to explain why you should use fragments, this is it.  The first three graphs should be in every introduction to FBHG. 
  • The Voldemort Rule: The Rule of Three paper.  This paper has defined what a fragment is for a decade. 
  • Fragments get a name (that's never used): Fragonomics.  OK, self-referencing is not cool, so this should really be Dan's paper, the first review on FBDD. 
  • Pfizer lifts the curtain: Pfizer's fragment library paper. I love this paper because it gives a great overview of how Big Pharma put its fragment library together (think laser pointers!). 
This is obviously a very short and incomplete list, and totally my opinion.  Let me know what you think MIP are in the comments.

So, what are "destructive" papers?  Those are the papers that require me to spend a lot of time explaining why what people understand is not really a good general, practical approach.  
Most Destructive Papers:
  • I think THE most destructive paper is also the most impactful: SAR by NMR.  How can that be you say?  Easy.  Because of this paper, the vast majority of people who "have heard" of fragments think that you need to label protein to do use NMR for fragments.  While, target-based screening is really powerful, I don't think it should be the first thought for screening, but is more impactful on active follow up.  I can here the counterarguments coming, but WAIT there's more.  They used linking, rather than growing.  I think most people would agree that this is the "Serendipity" approach.  I think this territory is well trod on this blog.  Lastly, warheads. 
  • While it had its place the Rule of Three paper has also become destructive.  This is a "hot" topic, but my main problem is the slavish devotion to an empirical "Rule".
What are your thoughts?  

I am also cross-posting this on my site http://www.quantumtessera.com/325/ and the LI group to see if maybe a different venue will generate more comments.

1 comment:

Dan Erlanson said...

First, thanks for the shout-out of the 2004 J. Med. Chem. paper! Although this wasn't actually the first review on fragment-based lead discovery, it was, together with the Astex Nat. Rev. Drug Disc. paper that same year, the most comprehensive review of the field to date.

The concept of destructive papers is interesting. The rule of three has been extensively discussed on Practical Fragments, and I promised no more posts on the topic for at least a month. However, the SAR by NMR paper has not received much attention on the blog, despite its seminal role in the field.

I do agree that, by focusing attention on linking, the paper did perhaps lead some folks down unproductive (and frustrating!) paths. That said, it remains a beautiful case-study even today, and linking can be useful, particularly for difficult systems.

As Teddy points out, though, the paper did firmly associate the notion of target-based NMR with FBLD, perhaps causing people to overlook the many other ways NMR can be used. More generally, I'd say that it has led people to assume that FBLD can only be done with structural information, an assumption with which I take issue.

Rod Hubbard often mentions the Gartner Hype cycle, and I think this is worth keeping in mind when discussing high-profile papers. The fact that the original SAR by NMR paper led to unrealistic expectations isn't really the fault of the paper, rather to the inevitable tendency to jump on The Next Big Thing. To the extent that the publication spurred people into the field, leading them to refine, alter, and perhaps even refute some of the initial ideas, I think it has actually been quite constructive.