Neural networks for improving drug discovery efficiency / Spring 2019 - Master Thesis
Antibiotic resistance represents one of the major threats to human health. The adaptation of pathogen requires a constant effort to find new drug and targets. The development of new drugs is becoming increasingly complex. To improve the efficiency and drive down cost, the drug discovery effort relies more and more on computational techniques to screen through large library of small molecules. However, few algorithms are currently able to correctly rank the most promising leads.
To overcome this problem, we propose to use machine learning for the prediction of target-ligand affinities. The datasets are composed of 3 dimensional structure of complexes. In this project, the candidate will have to identify the best representation(s) for encoding these complexes and key physicochemical properties that should be considered. A 3D-convolutional neural network will be trained and tested for prediction the absolute binding affinities.
Supervised by Thomas Lemmin
- Jimenez J, et al, KDEEP: Protein−Ligand Absolute Binding Affinity Prediction via 3D-Convolutional Neural Networks, JCIM, 2018
- P.J. Ballester & J.B.O. Mitchell, A machine learning approach to predicting protein-ligand binding affinity with applications t molecular docking, Bioinformatics, 2010