Computational prediction of the mechanisms of action of drugs using machine learning algorithms

Identifier:  TFG:4432

Handle: https://hdl.handle.net/20.500.11797/TFG4432

Authors:  Bel Bordes, Gemma

Abstract:
Over time, drug design costs have risen, although it has not impacted the number of successful drugs. Machine learning tools could potentially change this situation by introducing new methods to assess the properties of drugs, such as their mechanism of action. We have looked for the best model to predict these mechanisms, given gene expression and cell viability data of the cells treated with the drug, employing well-known algorithms such as support vector machines and a stochastic block model. The former resulted in the best predictive ability, while the latter remains a good model for interpreting the data interactions.