Desarrollo de modelos computacionales para identificar nuevos inhibidores de GAT-1

MANUEL COUYOUPETROU; ALAN TALEVI; MARÍA ESPERANZA RUIZ; GUIDO PESCE; LUIS BRUNO-BLANCH

Buenos Aires

Congreso; 8th. Latin American Congress on Epilepsy; 2014

INTERNATIONAL LEAGUE AGAINST EPILEPSY

GABA transporter 1 (GAT-1) is considered as the most important transporter for neuronal GABA uptake; it is a validated molecular target of antiepileptic medications, among them the clinical drug tiagabine [1]. The objective of this work was the development of an ensemble of computational models capable of identifying GAT-1 inhibitors. Methodology: A dataset of 106 GAT-1 inhibitors (IC5010μM) was compiled from bibliographic data. Such dataset was used to infer and validate a set of computational models capable of differentiating GAT-1 inhibitors and non-inhibitors. The models have been combined through different ensemble learning approaches. A simulated virtual screening campaign was performed on a simulated database containing less than 2% GAT-1 Inhibitors in order to estimate the ability of our model to retrieve GAT-1 inhibitors from large chemical libraries. Results: Receiving Operating Characteristic analysis showed that the modeling approach was successful in finding a model combination with high sensitivity (Se: 0.75) and specificity (Sp: 0.90). Conclusions: The ensemble developed is a useful tool to assist the computer-aided discovery of new drug candidates targeting GAT.1; these models are to be used in virtual screening campaigns to identify new anticonvulsant agents, in the near future