Descriptors Almond

I 9 Molecular Interaction Fields in ADME and Safety 9.3.1.3 ALMOND Descriptors... 

The structural variance of the dataset was analyzed with principal component analysis (PCA) [9] performed on the complete set of ALMOND descriptors calculated for the compounds which comprised the training and test sets. The first two components explained 35% of the stmctural variance of the dataset. Figure 9.1 shows that no structural outliers are present in the dataset and that the training and test sets share similar chemical space. 

Table9.1 Structural interpretation of the ALMOND descriptors that are highly correlated to the variance of the experimental data.

The 3D descriptors described in this section are the weighted holistic invariant molecular (WHIM) descriptors, the Jurs descriptors, and the GRID-based VolSurf and Almond descriptors, as well as pharmacophore fingerprints. 

Figure 10 Pairwise correlation between PCA scores derived from SaSA descriptors (2D) and Almond descriptors (3D). (See color plate at end of chapter.)...

D-molecular descriptors, alignment-independent and based on molecular interaction, called GRIND have been developed. These are autocorrelation transforms that are independent of the orientation of the molecules in 3D space. The original descriptors can be extracted from the autocorrelation transform with the ALMOND program. The basic idea is to compress the information present in 3D maps into a few 2D numerical descriptors which are very simple to understand and interpret. 

The GRIND were published in the year 2000, and the first version of the software used for generating and manipulating these descriptors (ALMOND) [7], was available in the same year. Since then, more than 20 applications of GRIND in the most diverse fields have been published (see Table 6.1). 

Einally, there is a fourth program which encapsulates GRIND descriptors named ALMOND. ALMOND theory is described in Chapter 6. What we will describe below refers to our involvement in utilizing the GRIND descriptors to model and understand various ADME and safety properties. 

The innovative autocorrelation algorithm used in ALMOND allows the representation of the descriptors in the original 3D space as a line linking two specific MIF nodes. 

The PLS multivariate data analysis of the training set was carried out on the descriptors matrix to correlate the complete set of variables with the activity data. From a total of 710 variables, 559 active variables remained after filtering descriptors with no variability by the ALMOND program. The PLS analysis resulted in four latent variables (LVs) with / = 0.76. The cross validation of the model using the leave-one-out (LOO) method yielded values of 0.72. As shown in Table 9.2, the GRIND descriptors 11-36, 44-49, 12-28, 13-42, 14-46, 24-46 and 34-45 were found to correlate with the inhibition activity in terms of high coefficients. 

PLS analysis was performed on the reduced set to identify a correlation between the complete set of variables and the activity data. The ALMOND program kept 624 active variables out of a total of 750 variables after filtering out descriptors with no variability. The 45 compounds with two basic nitrogen atoms were proton-ated on just one of the two basic centers. For these compounds, two models were built using one of the two isomers at a time. The isomer that produced the model with the best r was chosen. The PLS analysis resulted in a model with three LVs and r = 0.77. The cross validation of the model by the LOO method yielded q values of 0.74. As shown in Table 9.2, the GRIND descriptors that had high coefficients in the PLS model were 11-36, 44-58, 12-41, 13-40, 14-47, 24-52, and 34-49. 

All QSAR analysis was carried out using the program ALMOND 3.2.0 and grid alignment independent descriptors (GRIND). 

In addition to the VolSurf treatment of the GRID fields, the information from the MIF can also be transformed to obtain a pharmacophoric type of representation, which is useful in the modeling of metabolic stability, cytochrome inhibition or even the direct study of the ADME related proteins (Fig. 10.3). The Almond software [17] transforms the MIF into a distance-based representation of the molecule interaction. These parameters describe the geometry of the interaction and QSAR models can be derived where the interaction with a protein is essential. Detailed information on these descriptors is presented elsewhere in this book. 

These are molecular descriptors derived from molecular interaction fields (MIF) calculated by using different probes and representing the geometrical relationships among MIF regions [ALMOND - Multivariate Infometric Analysis s.r.L, 2007 Pastor, Cruciani et al., 2000 Cruciani, Pastor et al., 2001b Gratteri, Cruciani et al., 2001]. 

Finally, a burst of creativity from central Italy should be noted—the work of Pastor et al. [30] in the development of novel descriptors, GRIND, feeding into the novel model-building algorithms of ALMOND [31],... 

---