- GW Home
- About GW
- University Life
- News & Events
- Faculty And Staff
Rapid prediction of properties related to biological activity from NMR spectraTechnology #013-029-voutchkova
Questions about this technology? Ask a Technology Manager
Efficient determination of octanol-water coefficient (logP) is essential for predicting the ability of chemicals (drugs, cosmetics etc.) to enter the intended target region in the body. The octanol-water coefficient is also a property used by regulators for modeling toxicity of chemicals to humans and the environment. Current computational methods for logP prediction, require prior knowledge of exact chemical structure and are not applicable to mixtures, while experimental methods are inadequate for certain classes of chemicals, such as surfactants. A method of predicting logP before a structure is ascertained and that is applicable to mixtures will speed up the chemical evaluation process, saving time and money.
During synthesis of new chemicals, Nuclear Magnetic Resonance (NMR) spectroscopy data is routinely obtained. Using only this NMR data, the inventors’ software accurately determines logP. Critical features, such as carbon chain length and hydrocarbon saturation, are quantified from the NMR data. These features are then fed to a multivariate model that predicts logP of the NMR sample. The overall process is simple and can be incorporated into the regular compound synthesis workflow.
The inventors tested and optimized
the software with a training dataset consisting of diverse chemical structures.
Their model shows excellent correlation (R2= 0.956) between
experimentally obtained logP values and those determined using the software.
- Use NMR data to predict the ability of drugs, cosmetics or other commodity products to enter human body
- Model acute and chronic toxicity of synthetic chemical compounds
- Assess bioaccumulation of toxic chemicals in aquatic ecosystems
- Works without knowledge of chemical structure
- Only NMR data is necessary to caclculate logP
- Does not require extensive computational resources