Developing sensor activity relationships for the JPL electronic nose sensors using molecular modeling and QSAR techniques

We report a quantitative structure-activity relationship (QSAR) study using genetic function approximation (GFA) to describe the polymer-carbon composite sensor activities in the JPL electronic nose (ENose), when exposed to chemical vapors at parts-per-million (ppm) concentration levels. A unique QSAR molecular descriptor set developed in this work combines the default analyte property set (thermodynamic, structural etc.) with sensing film-analyte interactions that describes the sensor response. These descriptors are calculated using semi-empirical and molecular modeling tools. The QSAR training data set consists of 15-20 analyte molecules specified by NASA for applications related to life support and habitation in space. The statistically validated QSAR model was also tested independently to predict the sensor activities for test analytes not considered in the training set