New predictive models for blood-brain barrier permeability of drug-like molecules

Abstract

Purpose: The goals of the present study were to apply a generalized regression model and support vector machine (SVM) models with Shape Signatures descriptors, to the domain of blood-brain barrier (BBB) modeling.

Materials and methods: The Shape Signatures method is a novel computational tool that was used to generate molecular descriptors utilized with the SVM classification technique with various BBB datasets. For comparison purposes we have created a generalized linear regression model with eight MOE descriptors and these same descriptors were also used to create SVM models.

Results: The generalized regression model was tested on 100 molecules not in the model and resulted in a correlation r2 = 0.65. SVM models with MOE descriptors were superior to regression models, while Shape Signatures SVM models were comparable or better than those with MOE descriptors. The best 2D shape signature models had 10-fold cross validation prediction accuracy between 80-83% and leave-20%-out testing prediction accuracy between 80-82% as well as correctly predicting 84% of BBB+ compounds (n = 95) in an external database of drugs.

Conclusions: Our data indicate that Shape Signatures descriptors can be used with SVM and these models may have utility for predicting blood-brain barrier permeation in drug discovery.

Fig. 1

1D and 2D Shape Signatures of fluoxetine (BBB+). a Chemical structure. b 1D (shape only) signature histogram. c 2D (shape and polarity) signature plot.

Fig. 2

Correlation of predicted logBB values (x-axis) versus the experimental logBB values (y-axis) for compounds from Xu-training set. The regression equation model resulted in an r²=0.70.

Fig. 3

a Results of the PCA analysis on the Xu–Combined dataset conducted in the space of eight molecular descriptors computed with MOE (PC1=54%, PC2=26%, PC3=10%). b Results of the PCA analysis performed on the Xu–Li dataset conducted in the space of eight molecular descriptors computed with MOE (PC1=52%, PC2=27%, PC3=11%). Black circles: molecules from Xu’s dataset. Red circles: BBB+ compounds from Combined (a) and Li’s (b) datasets. Blue circles: BBB- compounds from Combined (a) and Li’s (b) datasets.

Fig. 4

Results of the PCA analysis conducted in the space of 2D Shape Signatures (shape + charges) molecular descriptors on the Combined—SCUT and Li—SCUT datasets. a PC1 vs PC2 for the Combined SCUT dataset (PC1=55%, PC2=12%, PC3=9%). Black circles: 351 compounds from Combined dataset. Red circles: 95 BBB+ compounds from SCUT. Blue circles: 294 BBB- compounds from SCUT. b PC1 vs PC2 for the Li-SCUT dataset (PC1=63%, PC2=11%, PC3=8%). Black circles: 378 compounds from Li dataset. Red circles: 95 BBB+ compounds from SCUT. Blue circles: 294 BBB- compounds from SCUT.