CORAL: QSAR models of CB1 cannabinoid receptor inhibitors based on local and global SMILES attributes with the index of ideality of correlation and the correlation contradiction index

Obesity has acquired notable attention due to its high occurrence and link with grievous health problems such as hypertension, diabetes and heart disease. It has been reported that the endocannabinoid system executes a pivotal part in the management of food absorption, fa augmentation, and energy balance. In the present manuscript, we report a detailed QSAR analysis for 165 CB1 cannabinoid receptor inhibitors employing the Monte Carlo optimization process incorporated within the CORAL software. Eight splits are made from the whole dataset and sixteen QSAR models are developed from these splits employing two target function TF1 (without index of ideality of correlation) and TF2 (with index of ideality of correlation). All the QSAR models developed with TF2 have better predictive potential than the models developed with TF1. The model built for split 5 using TF2 is the leading model due to the higher value of the determination coefficient of the validation set (R-valid(2) = 0.8518). The index of ideality of correlation (IIC) improves the statistical performance of CORAL-based QSAR-models and gives statistically robust predictive models of the investigated endpoint pIC(50). In the present manuscript, a novel criterion Correlation Contradiction Index (CCI) is also applied to know its predictive potential. The absolute value of CCI for calibration set is less when QSAR models are developed employing IIC. The promoters of increase and decrease endpoint pIC(50) are identified and these are applied to design seven new compounds. All the newly designed molecule were docked into in the active site of human cannabinoid receptor CB1 (PDB ID: 5tgz).