Descriptor alerts

Of course, class 4 is not valuable in medicinal chemistry. Such compounds have to be excluded from drug discovery processes as early as possible. At present, there are computer alert programs based on the Rule-of-5 or similar approaches that are used in preliminary screening to select and exclude compounds of class 4 [71]. Van de Waterbeemd indicated in 1998 that the four BCS classes of drugs can be determined solely by considering physicochemical descriptors such as molecular weight and PSA [72]. However, as mentioned in this chapter, those descriptors are too crude for the quantitative description of molecular size and H-bonding ability. 

All the descriptors and properties necessary to calculate J-Alert are generated by the software program QMPRPlus (Simulations Plus, Inc.). The current set of QMPRPlus computational models for biopharmaceutical properties is listed below ... 

One of the simplest and most common ways to evaluate a molecule for ADME properties is a qualitative examination of its basic descriptor values such as molecular weight (MW), ClogP for lipophilicity, polar surface area (PSA), counts of hydrogen bond donors and acceptors (HBD, HBA), and count of rotatable bonds (RB). This type of approach popularized by Lipinski s famous Rule of 5 was published a decade ago [6]. Lipinski et al. established cutoffs for MW (500), ClogP (5), HBA (10), and HBD (5). These cutoffs were based on the 90th percentile of distributions of molecules in the World Drug Index having USAN or INN names. The Rule of 5 considers a violation of any two of these cutoffs to be an alert for poor absorption or permeability. 

We have also undertaken an evaluation of this and other datasets using a simple descriptor analysis as well as readily available substructure alerts or filters (15-17). The GSK, St Jude, and Novartis datasets also have very high failure rates with the Abbott Alerts (18, 19) (75-85%) and Pfizer Lint filters (40-57%) (Fig. 4). A set of 14 US Food and Drug Administration (FDA) approved... 

An exhaustive analysis of 2995 molecule pairs extracted from the 98.1 version of Bioster database indicated that similarity measures based on 2D molecular fingerprints or electrostatic field descriptors were complementary although 2D methods could be adequate for similarity analyses [55]. To evaluate a range of similarity measures among synthetic substances and natural products, the Willett group also used 5024 compounds from Bioster database as well as sets of selected bioactive compounds from the more populous Chemical Abstract Service, ID-Alert, MACCS Drug Data Report, and NCI AIDS databases [56]. 

With this, we would like to end the introduction into justification for searching for novel molecular descriptors specifically designed for searching combinatorial libraries and highly similar structures. We hope that this may alert researchers looking for novel drugs to discuss this problem and, despite a few additional applications. 

Structural alert QSARs for the trial compounds of Table 3.36 gave the stmctural alert activities shown in Table 3.40. Full molecular QSARs for the trial molecules of Table 3.36 are reported in Table 3.41 as M computed/predicted models. Combined QSAR predictions based on the molecular descriptors fi om Table 3.36 and the structural alert activities of Table 3.40 are reported in Table 3.41 as MaS4 these results showcase... 

To this aim one may return to the structural-alert analysis and employs the euclidean paths for residual-alert OECD-QSARs of step (ii-bis) (Table 3.43) for the trial molecules of Table 3.36 and the test compoimds of Table 3.37. The models were arranged so that each model emerges from the previous one on the basis of their common descriptors the results are reported in Table 3.45 by employing the Euclidean path between two successive QSAR models (eomputed endpoints), Eq. (3.125). The optimum paths for residual-alert QSARs are derived by searehing the minimum paths and the assoeiated hierarchy according to the formal constraint of Eq. (3.126) (Putz Lacrama, 2007 Lacr a et al., 2007 Putz et al., 2009b, 2010 Chicu Putz, 2009) for the residual-QSAR regression (endpoint) models computed with 1, k,. .., mstructural parameters (under the OECD-QSAR step (v)), as prescribed by... 

---