Inhibition of the
matrix metalloproteinases (
MMPs) is effective against
metastasis of secondary tumours. Previous
MMP inhibitors have failed in clinical trials due to their off-target toxicity in solid tumours. Thus, newer
MMP inhibitors now have paramount importance. Here, different molecular modelling techniques were applied on a dataset of 110
gelatinase (MMP-2 and MMP-9) inhibitors. The objectives of the present study were to identify structural fingerprints for
gelatinase inhibition and also to develop statistically validated QSAR models for the screening and prediction of different derivatives as MMP-2 (
gelatinase A) and MMP-9 (
gelatinase B) inhibitors. The Bayesian classification study provided the ROC values for the training set of 0.837 and 0.815 for MMP-2 and MMP-9, respectively. The linear model also produced the leave-one-out cross-validated Q2 of 0.805 (eq. 1,
MMP-2) and 0.724 (eq. 2,
MMP-9), an r2 of 0.845 (eq. 1,
MMP-2) and 0.782 (eq. 2,
MMP-9), an r2Pred of 0.806 (eq. 1,
MMP-2) and 0.732 (eq. 2,
MMP-9). Similarly, non-linear learning models were also statistically significant and reliable. Overall, this study may help in the rational design of newer compounds with higher
gelatinase inhibition to fight against both primary and secondary
cancers in future.