Receptor Binding Affinity of Dibenzofurans

For comparative purposes, the same data set was analyzed using an alternative approach. Rather than employing the hierarchical method, we began with the entire pool of descriptors, viz., TS + TC + 3D + STO-3C, and performed variable thinning via the modified Cram-Schmidt algorithm. RR, 

and PLS regression analyses were performed using the reduced set of 72 descriptors. Again, a high-quahty RR model was obtained with a value of 0.899. Comparing the descriptors with the highest t -values in this model and those in the best RR model obtained using the hierarchical approach, i.e., TS + TC, it was foimd that the same descriptors are important in both models (Table 5). 

A perusal of the descriptors important for the correlation of Ah receptor binding affinity (Table 5) shows that electronic factors, as well as general molecular shape and size, play a dominant role. The SaaCH, SHarom, and siunDell indices encode information about the hybridization and electronic states of atoms in the molecule. The x , Idcbar and P4 descriptors are quantifiers of general molecular shape and size and are probably related to steric fit between the receptor and the ligand. In our PCA study involv- 

Chemical Experimental pECso Predicted pECso Exp. - Pred. 

Triplet index from adjacency index, graph order, and vertex degree operation = summation 

---

Tables Descriptors important for the prediction of Ah receptor binding affinity of dibenzofurans, identified in both the hierarchical TS + TC RR model and the RR model developed following Gram-Schmidt variable thinning...

---