Application of artificial intelligence in drug discovery
Development, validation and application of machine learning and deep learning algorithms for the automated generation (de-novo design) of bioactive compounds.
This research activity aims at developing, validating and applying new generative algorithms based on artificial intelligence (machine learning and deep learning) in order to minimize costs, time and environmental impact typically required for developing new drug candidates. To achieve this aim, two steps are followed: i) the learning phase, taking advantage of the information provided by a large repository of bioactive data (such as for instance ChEMBL); ii) the generation phase, performed thanks to the application of recurrent neural networks (RNN) or variational autoencoders. The quality of the generated compounds is challenged by structure-based methodologies (molecular docking) and the best candidates tested in-vitro.
– Creanza, T.M., Lamanna, G., Delre, P., Contino, M., Corriero, N., Saviano, M., Mangiatordi, G.F., Ancona N., DeLA-Drug: A Deep Learning Algorithm for Automated Design of Druglike Analogues (2022) Journal of Chemical Information and Modeling 62 (6), 1411-1424.
– Montaruli, M., Alberga, D., Ciriaco, F., Trisciuzzi, D., Tondo, A.R., Mangiatordi, G.F., Nicolotti, O. Accelerating drug discovery by early protein drug target prediction based on a multi-fingerprint similarity search † (2019) Molecules, 24 (12), art. no. 2233, .
– Alberga, D., Trisciuzzi D., Montaruli, M., Leonetti, F., Mangiatordi, G.F., Nicolotti, O. A New Approach for Drug Target and Bioactivity Prediction: The Multifingerprint Similarity Search Algorithm (MuSSeL) (2019) Journal of Chemical Information and Modeling, 59 (1), pp. 586-596.