Fatty Acids Fingerprint Profiling And Authenticity Modeling Of Animal And Vegetable Oils

Animal and vegetable oils are very diverse and widely available in prices,and high value meat,milk and oils are highly susceptible to adulteration with cheap products.At present,there is a lack of systematic research on fatty acid(FAs)fingerprint profiling and authenticity discrimination model covering animal and vegetable oils.137 samples of 5kinds of livestock meat,137 samples of 6 kinds of livestock milk,121 samples of 8 kinds of vegetable oil and 16 samples of waste oil(12 samples of kitchen waste oil and 4 gutter oil)were collected,totaling 411 samples of 21 kinds of oil in 4 categories.Gas chromatography was used to detect FAs,and descriptive statistics were made for each classification.Principal component analysis(PCA),orthogonal partial least squares discriminant analysis(OPLS-DA)and hierarchical cluster analysis(HCA)were used to observe the characteristics of natural clustering and supervised clustering of various oil samples,and to evaluate the pattern similarity of each classification.OPLS-DA and soft independent modeling classification(SIMCA)discriminant models were constructed to evaluate the feasibility of FAs to identify the authenticity of animal and vegetable oil species.PCA and OPLS-DA can cluster and separate all kinds of oil,which accords with the law of species taxonomy.OPLS-DA was obviously better than PCA.The pattern similarity of HCA was consistent with the clustering results.The cluster of meat and milk fat was close,and the pattern similarity was 54.76%,while the vegetable oil was far away as an independent cluster.Ruminant meat and milk clusters were close to each other,bactrian camels were far away,and monogastric horses and pigs ware far away.The milk pattern similarity of bovine,goat,sheep and yak was 88.99%,sheep and goat milk was99.06%,yak and cow milk was 91.48%,horse and other milk was 60.50%.Camel milk fell into meat cluster,the pattern similarity of FAs with bactrian camel meat was 97.24%.All kinds of vegetable oils were clustered separately.Peanut oil and flaxseed oil have unique FAs fingerprint patterns,and the similarities with other vegetable oil patterns were70.25% and 84.79% at the left and right ends of the vegetable oil clustering area,respectively.Blended oil crosses soybean,corn,sunflower and sesame oil.87.5% of waste oils fall into vegetable oil clusters,and the rest fall into meat.According to binary analysis,waste oils can be separated from them.The external verification accuracy of OPLS-DA and SIMCA models for meat,milk,vegetable oil and waste oil was 94.6% and 95.8% respectively.The external verification accuracy of meat model for 5 kinds of livestock meat samples was 98.5% and 94.0%respectively.The accuracy of external verification of milk model for 6 kinds of milk were100% and 89.3% respectively.In the oil model,except blended oil,the external verification accuracy of 7 vegetable oils was 100% and 89.4% respectively.The accuracy of distinguishing waste oil from vegetable oil was 75.0% and 57.5%,and the accuracy of distinguishing waste oil from meat fat was 85.0% and 82.5%.The model of distinguishing waste oil needs more attempts to optimize.The overall OPLS-DA discrimination effect was better than SIMCA.It is feasible to identify the authenticity of animal and vegetable oil species by FAs fingerprint profiling,which provides innovative ideas and methods for the authenticity identification of animal and vegetable oil species.