6533b7dbfe1ef96bd126fefb
RESEARCH PRODUCT
Biodegradability Prediction of Fragrant Molecules by Molecular Topology
Vincent BlayRamón García-domenechMaria Galvez-llompartJorge GalvezJesús Gullón-soletosubject
Artificial neural network010405 organic chemistryRenewable Energy Sustainability and the EnvironmentComputer scienceStatistical learningGeneral Chemical EngineeringNanotechnologyLinear classifierGeneral Chemistry01 natural sciences0104 chemical sciencesCost reduction010404 medicinal & biomolecular chemistryDevelopment (topology)SAFEREnvironmental ChemistryBiodegradability predictionBiochemical engineeringMolecular topologydescription
Biodegradability is a key property in the development of safer fragrances. In this work we present a green methodology for its preliminary assessment. The structure of various fragrant molecules is characterized by computing a large set of topological indices. Those relevant to biodegradability are selected by means of a hybrid stepwise selection method to build a linear classifier. This model is compared with a more complex artificial neural network trained with the indices previously found. After validation, the models show promise for time and cost reduction in the development of new, safer fragrances. The methodology presented could easily be adapted to many quasi-big data problems in R&D environments.
| year | journal | country | edition | language |
|---|---|---|---|---|
| 2016-07-06 | ACS Sustainable Chemistry & Engineering |