Characterization of volatile flavour compounds and characteristic flavour precursors in poultry eggs based on multi-omics and machine learning

Abstract

Although egg flavour influences consumer preference and satisfaction, the features governing the flavour profiles have been poorly studied. This study investigated the volatile compounds and lipid profiles of different poultry egg yolks (chicken, duck, goose, quail and pigeon) using E-nose, volatile metabolomics and untargeted lipidomics, while also exploring the relationships between these components. Additionally, machine learning models (gaussian naive bayes, logistic regression, random forest, SVM and XGBoost) were developed to classify poultry egg types and identify biomarkers with high predictive accuracy. Overall, 11 key flavour compounds were identified as significant contributors to the flavour of poultry eggs. Analysis of flavour precursors indicated that the key flavour compounds were primarily originated from unsaturated fatty acids (UFAs) and PE, PC, TAG and ACAr containing UFAs. Furthermore, 9 markers (4 volatile compounds and 5 lipid metabolites) for distinguishing different eggs were obtained by constructing machine learning models.