Prediction of the heats of combustion for food-related organic compounds. A quantitative structure–property relationship (QSPR) study

Abstract

In the field of food research, the determination of the heats of combustion (Δ_cH) of the nutrients is essential to estimate the amount of energy obtained by metabolizing during digestion. Here, we have developed six novel QSPR models to predict this thermodynamic property of different families of organic compounds. The models were developed using the experimental data set of 215 compounds (71 organic acids, 28 amino acids, 37 amines and amides, 31 sulfur compounds and 48 heterocyclic compounds). About 16,000 molecular descriptors were calculated to represent the molecular structure of the compounds. The QSPR models resulted to be simple MLRs with a maximum of three variables, facilitating the interpretation and comparison with existing models in the literature. The statistical parameters exhibited excellent predictive capacity and robustness of the models obtained. The correlation coefficients of the selected models were major to 0.8 and the root means square error minor to 0.1. These results suggested that the models could be utilized for the prediction of the Δ_cH of other compounds that could be present in the foods.