Drug-likeness prediction

Fig. 17.6 Distribution of predicted drug-like scores for the test sets of ACD (dashed line) and WDI (continuous line) compounds.

Table 2.2 Em pi rical rules used to predict drug-likeness.

The focus of this book is on methods and processes designed to predict drug-like properties, exposure and safety during hit and lead discovery. We do not intend to cover specific cultural considerations and marketing aspects [3]. What we will highlight is the need of a risk aware environment for drug discovery, where data-based integrated risk assessment is part of daily life of the team and drives the projects towards molecules with features fit for the description of an efficacious and safe medicine. 

Table 13.2 Rank order of virtual combinatorial libraries based on predicted drug-likeness, cytotoxicity, GPCR-ligand likeness, and kinase-ligand likeness. ++ indicates pronounced positive prediction, + mediocre to slightly positive prediction, - negative prediction (absence of the property). 

Table 11.1 Lipinski parameters (R05) for predicting "drug-likeness" of potential molecular targets...

Prediction of various physicochemical properties such as solubihty, lipophhicity log P, pfQ, number of H-donor and acceptor atoms, number of rotatable bonds, polar surface area), drug-likeness, lead-likeness, and pharmacokinetic properties (ADMET profile). These properties can be applied as a filter in the prescreening step in virtual screening. 

Most practical implementations of drug-likeness use a computational model which takes as input the molecular structure, together with various properties, and predicts whether the molecule is drug-like or not. Some of these models may be very simple, such as a series of substructural filters. Only those molecules which pass all of these filters are output, Such filters can be used to eliminate molecules that contain inappropriate functionality. 

Walters WP, Murcko MA (2002) Prediction of drug-likeness. Adv Drug Deliv Rev 54(3) 25 5-271... 

Clark DE, Pickett SD. Computational methods for the prediction of drug-likeness . Drug Discov Today 2000 5 49-58. 

Zernov VV, Balakin KV, Ivaschenko AA, Savchuk NP, Pletnev IV. Drug discovery using support vector machines. The case studies of drug-likeness, agrochemical-likeness, and enzyme inhibition predictions. J Chem Inf Comput Sci 2003 43(6) 2048-56. 

From an analysis of the key properties of compounds in the World Dmg Index the now well accepted Rule-of-5 has been derived [25, 26]. It was concluded that compounds are most Hkely to have poor absorption when MW>500, calculated octanol-water partition coefficient Clog P>5, number of H-bond donors >5 and number of H-bond acceptors >10. Computation of these properties is now available as a simple but efficient ADME screen in commercial software. The Rule-of-5 should be seen as a qualitative absorption/permeabiHty predictor [43], rather than a quantitative predictor [140]. The Rule-of-5 is not predictive for bioavail-abihty as sometimes mistakenly is assumed. An important factor for bioavailabihty in addition to absorption is liver first-pass effect (metaboHsm). The property distribution in drug-related chemical databases has been studied as another approach to understand drug-likeness [141, 142]. 

Blake, J. F. Chemoinformatics -predicting the physicochemical properties of drug-like molecules. Curr. Opin. Biotechnol. 2000, 11, 104-107. 

Many excellent computer programs are available for predicting log P a from two-dimensional structures. The quality of predictions has risen over the years to the point that rouhne log P a measurements are not regularly done at some pharmaceutical companies, but rather, calculated values are used. It is worth noting that log Port values of newly synthesized classes of drug-like compounds sometimes are still poorly predicted and probably there will be the need for judicious log Port measurements for years to come. 

Balakin, K., Savchuk, N., Tetko, I. In silica approaches to prediction of aqueous and DM SO solubility of drug-like compounds trends, problems and solutions. Curr. Med. Chem. 2006, 13, 223-241. 

Delaney [4,14] and Klamt [16] argued that for drug-like compound datasets only about 20% of the variance of log S arises from AG s. This is further confirmed by the study of Wassvik et al. [15] in which 77% of the variance is due to the solubility of the supercooled liquid. Hence, applying crude estimates by mean values or by QSAR approaches we can reasonably expect that the inaccuracies introduced in dmg solubility prediction by our theoretical ignorance of AG s is less than, or at least not much bigger than, the inaccuracies introduced by the estimates of the larger park i.e. the liquid solubility, and by the experimental difficulties in solubility measurement. 

---