Matched molecular pair analyses

An alternative approach to analysis of structural databases is to focus on relationships between structures. A relationship between structures can be a simple substitution (6.21a is mono-chloro analogue of 6.21b) or something more abstract (6.22a and 6.22b are reversed amides). 

---

Tab. n.5 Matched molecular pair analysis of the effect of substituent on aqueous solubility... 

In what follows, the working definitions of MMPs, transformations, and contexts will be given. A brief review of the related literature on the subject of matched molecular pair analysis vfill follow. Finally, the last section ofthis chapter will focus on a large-scale experimental study that was carried out. Therein, a detailed account of the experiments will be given and the findings will be discussed. 

Table 6.1 Recent studies on matched molecular pair analysis.

Figure 10.8 Histograms showing distribution of change in iog(potency) for isosteric replacement of 3-phenoi, as derived from a matched molecular pair analysis of the Pfizer IC50 data set. In addition, the ratio of replacements that are as good (at ieast within...

Figure 11.4 Assessment of Drug Guru performance of the Remove -Cl transformation for reducing hERG affinity (dofetilide displacement assay) using a matched molecular pair analysis of 151 compound pairs.

Birch AM, Kenny PW, Simpson I, et al. Matched molecular pair analysis of activity and properties of glycogen phosphorylase inhibitors. Bioorg Med Chem Lett 2009 19 850-853. 

Matched molecular pair analysis in drug discovery. Drug Discovery Today, 18 (15-16), 724-731. 

Schematic example of a Matched Molecular Pair Analysis (MMPA)...

Eor a brief analysis of what adding a 2-Cl substituent to a biphenyl system can do to bond angles and solubility, see Leach, A.G. Matched molecular pairs as a guide in the optimization of pharmaceutical properties A Study of aqueous solubility, plasma protein binding, and oral exposure. J. Med. Chem. 2006, 49, 6672-6682, and supplementary information thereto. 

More recently, Hajduk and Sauer analyzed the molecular transformations in Abbott Laboratories in-house data (84 000 structures) and their effect on the potency against 30 proteins [23]. They generated the matched molecular pairs using an algorithm based on the MCS function provided by the Daylight Toolkit Their analysis was limited to chemically sensible transformations, such as group additions (i.e., H 3> Y) and multiple region-specific phenyl substitutions. The results illustrated that no transformation seemed to consistently favor potency increase. 

The molecular recognition between mismatched (R,R)-29 and (S,S)-43 is less efficient compared with the matched pair complex. Even if the complex (R,R)-29 (S,S)-42 could be obtained in a crystalline form (Scheme 19), the X-ray analysis shows that one H-bond between the hydroxy and carboxylate moieties is lost and coordination has been achieved with two water molecules included in the crystal (Figure 52). The loss of such water molecules, for example after drying of the sample, determines the erosion of crystallinity. 

The principle concern of MMPs for chemical structures is how the trade-off between specificity and general trend is achieved [66]. A very specifically defined environment for a chemical transformation results in a much narrower dataset for statistical analysis. The extracted trend might then be clearer for this particular problem, but based on a smaller dataset. Hence, any further generalization might be difficult. If on the other hand, the molecular context is completely nnconstrained, then the effect of a particular substitution is likely to reflect only simple trends, mainly related to lipophilicity as one of the cardinal properties in medicinal chemistry [62, 66, 67]. The simpler the property for MMP analysis is, the more likely is a direct link to molecular structure, as shown by Leach et al. in their study on solnbihty and plasma protein binding [62]. Typical control parameters for stable MMP trends are thus related to a minimum number of matched pairs for analysis and a certain degree of structural diversity upon all members in a dataset, excluding of conrse those linked by a molecular transformation [71]. 

The analysis for this limit is very similar to that of the weak-coupling limit, except that now the hardcore repulsion imposes the boundary condition that V (O) = 0 on all the solutions. Thus, degenerate pairs of even and odd solutions are found by matching (r) with tp —r) as r —> 0. So, setting the molecular-orbital... 

---