Quantitative Structure-Property Relationships for Predicting the Retention Indices of Fragrances on Stationary Phases of Different Polarity


Abstract:

The purpose of this work was to develop predictive quantitative structure–property relationships for modeling the retention indices (I) of fragrances measured in three stationary phases of different polarities: DB–225MS, HP5–MS and HP–1. Attention was paid to the curation of the experimental data. Subsequently, the Balanced Subsets method (BSM) was used to split each dataset into training, validation and test sets. Models were established by using 1819 conformation–independent molecular descriptors which were analyzed by the replacement method (RM) variable subset selection in order to obtain the optimal models. A four–descriptor model was obtained for the DB–225MS stationary phase while a three–parametric model was proposed for both the HP5–MS and HP–1 columns. Models were validated by means of the leave–one–out and leave–many–out cross–validation procedures, as well as other validation criteria. Moreover, in order to accomplish the principles proposed by the Organization for Economic Co–operation and Development (OECD), the model’s predictive ability was measured by predicting retention indices of the external test set. The applicability domain was properly defined and the interpretation of each of the molecular descriptors used in this study was provided.

Año de publicación:

2017

Keywords:

    Fuente:

    googlegoogle

    Tipo de documento:

    Other

    Estado:

    Acceso abierto

    Áreas de conocimiento:

    • Relación cuantitativa estructura-actividad
    • Ingeniería química
    • Ciencia de materiales

    Áreas temáticas de Dewey:

    • Química física
    • Química analítica
    Procesado con IAProcesado con IA

    Objetivos de Desarrollo Sostenible:

    • ODS 9: Industria, innovación e infraestructura
    • ODS 12: Producción y consumo responsables
    • ODS 3: Salud y bienestar
    Procesado con IAProcesado con IA

    Contribuidores: