Steps, calculations and results of studies of theoretical predictions of quantitative structure retention relationship (QSRR) of Pittosporum undulatum essential oil

Document Type : Original Article

Authors

• M M
• M N

IAU

Abstract

In this article, a detailed description of the linear models capable of predicting the inhibition indices of a large group of natural compounds identified in the essential oil of Pittosporum undulatum, as one of the medicinal plants, has been discussed. In this regard, the work is based on quantitative structure retention relationship (QSRR), which is of prime importance in scientific resources to establish a logical and meaningful relationship between the Kovats index as a dependent variable and a group of molecular descriptors as independent variables. In this regard, after drawing the structure of the natural compounds using the Hypercam software and optimizing their molecular structures, Dragon software was used to extract the relevant molecular descriptors. In the next step, after removing irrelevant and redundant descriptors, a group of important and effective descriptors were identified and their linear relationship with the Kovats retention index was discussed and investigated using stepwise multiple linear regression method as well as another variable selection method based upon genetic algorithm feature selection approach. The obtained results indicate the high capability of the presented models to predict the Kovats index of a wide group of natural compounds.

Keywords

• Quantitative structure retent relationship (QSRR)
• multiple linear regression
• molecular descriptors
• genetic algorithm
• Kovats retention index
• Pittosporum undulatum

---