Wednesday, 18 February 2009

Substituents and complexity

<< previous || next >>

In the previous post, I noted that two Astex kinase inhibitors were derived from fragments that lacked acyclic substituents. Dan points out that this is actually uncommon and wonders if this reflects a reluctance of medicinal chemists to work on fragments that were seen to be too simple.

The presence of certain molecular recognition elements, for example hydroxyl or carboxylate, implies that at least one acyclic substituent be present. I think this it probably the main reason that fragments are normally encountered with acyclic substituents. However, I do agree with Dan that some fragments can be seen as too simple and re-iterate my point that in the Brave New World of FBDD we really need to start seeing phenyl rings as synthetic handles.

A lack of acyclic substituents typically implies the presence of one or more polar atoms in a ring or spacer. When assembling screening libraries, I do try to select compounds that present heterocyclic molecular recognition elements without acyclic substituents (e.g. 4-phenypyrazole, 2-anilinopyrimidine). Interestingly compounds like these are not as easy to source as you might think.

Saturday, 14 February 2009

Molecular complexity and extent of substitution

<< previous || next >>

Having introduced extent of substitution as a measure of molecular complexity in an earlier post, I was particularly interested by Dan's posts on AT7519 and AT9283. In each case, the screening hit used as a starting point for further elaboration lacked acyclic substituents.

You might wonder how you could impose this substructural requirement when selecting compounds for screening. This is actually very easy using SMARTS notation (Daylight SMARTS tutorial | OpenEye SMARTS pattern matching | SMARTS in wikipedia). The requirement that terminal non-hydrogen atoms be absent can be specified as:

[A;D1] 0

D1 indicates a non-hydrogen atom (A) that is connected to only one other non-hydrogen atom and 0 requires that these cannot be present in acceptable molecules. A requirement like this can be combined with a requirement for 10 to 20 non-hydrogen atoms:

* 10-20

I will discuss the use of SMARTS for compound selection in more detail in connection with design of screening libraries so think of this as a taster. I've also tried to keep things simple by assuming that hydrogen atoms are implicit which means that they are treated as a property of the atoms to which they are bonded rather than as atoms in their own right.

Tuesday, 3 February 2009

Ligand efficiency and molecular size

<< previous || next >>

Molecular complexity models of ligand binding typically predict that ligand efficiency (LE) will decrease during the process of elaborating a fragment hit. This is the basis of the fit quality (FQ) metric that Mel reviewed in her last post. It’s interesting to take a look at a 2006 article which actually pre-dates the FQ articles. This study attempted to track LE through the lead optimisation process for 18 drug leads from 15 different projects. The main conclusion of the study was that ‘a nearly linear relationship exists between molecular weight and binding affinity over the entire range of sizes and potencies represented in the dataset’. In other words, LE changed little during the optimisation process for these leads.

Comments anyone?

Literature cited:
Hajduk, Fragment-Based Drug Design: How Big Is Too Big? J. Med. Chem. 2006, 49, 6972-6976 DOI