Analysis of machine learning methods that allow the generation of novel molecules with desired properties using existing databases of known molecular compounds
The main purpose of this project is to facilitate drug discovery by providing means of generating a diversity of novel molecules in silico. While convenient methods of drug discovery involve long cycles of synthesis and testing molecules in vitro, the generational machine learning techniques aim to eliminate some of the most expensive and time-consuming stages substituting them with cheap and fast generation using machine learning. The objective of my project is about the application of machine learning algorithms to generate de novo molecules with desired properties. To achieve such aim I am using an existent database of commercially available compounds. I would like to acknowledge the support of NPO Young Researchers Alliance and Nazarbayev University Corporate Fund “Social Development Fund” for grant under their Fostering Research and Innovation Potential Program.