Predicting molecular properties such as the lipophilicity of a drug molecule is crucial for the drug discovery field. Lipophilicity (logP) is the solubility of the drug in the lipid membrane, which is related to the partition coefficient of the molecule in a two-phase system such as water and octanol. Despite recent advancements in the bioanalysis field for measuring logP, experimental logP values for most molecules are still not available. These values could be estimated using computational methods such as quantum mechanics and machine learning.<br/><br/>In this project, we use the available logP data set of 4000 drug-like molecules, and then we generate new 3D conformers for each molecule. These 3D structures were used then as input for DFT calculations in the gas, 1-octanol, and water phases. More than 50 molecular descriptors were parsed from these simulations. The resulting molecular descriptors are then used to train a machine learning model such as random forest to predict the logP values.

Computational prediction of lipophilicity of the molecule: Quantum chemistry and machine learning approach

ACS Spring Meeting 2021

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.