Validating the ability of REIMS to differentiate lamb flavor performance based on consumer preference

The objective of this study was to investigate the capability of rapid evaporative ionization mass spectrometry (REIMS) to accurately identify and predict cooked sheep meat flavor and carcass characteristics based on consumer response utilizing metabolomics data acquired from raw samples. A total of 200 boneless leg samples were collected from sheep representing two age classifications (n = 99 lamb, n =101 yearling), at three USDA harvest facilities located in California and Colorado. Collections were completed between the months of September 2019 to January 2020. Consumer sensory panels consisted of 200 panelists who answered a series of questions regarding flavors and overall liking of ground sheep patties. The REIMS platform captured metabolomic data from lean tissue, external fat tissue and ground sensory patties. New binary values were developed as well as levels of intensity based on consumer responses for the development of predictive models. Principal component analysis (PCA) was utilized to reduce dimensionality of the data prior to creating the model using linear discriminant analysis (LDA). Additionally, top 100 REIMS bins were determined by an f-statistic value and applied to reduce dimensionality before creating the model using LDA. Moreover, top 100 REIMS bins were entered into UCSD Metabolomics Workbench to determine possible compounds responsible for sheep characteristics and flavor attributes (Metabolomics Workbench, University of California San Diego, San Diego, California, USA). Results show selection from the top 100 REIMS bins were more effective overall when classifying attributes and predicting overall accuracy. Overall accuracies were above 80% for all sheep flavor attributes and characteristics. Production background (grain-fed or grass-fed) revealed the highest classification accuracies with the lean tissue and ground sensory patties were classified with 100% accuracy while external fat tissue was 99.50% accurate. Off-flavors were identified external fat tissue with the highest classification accuracy at 92.64%, lean tissue at 84.66% and ground sensory patty at 86.50%. While the PCA-LDA model classified off-flavor at most with a 72.34% overall accuracy obtained by the external fat tissue. The PCA-LDA showed lower overall accuracies for most all sheep flavors and characteristics in comparison to the top REIMS 100 bins model. However, the PCA-LDA model classified production background (grain-fed or grass-fed) at high overall accuracies with lean tissue at 97.50%, external fat tissue at 95.00% and ground sensory patty at 97.50%. Moreover, promising results were discovered from the metabolomics database identifying possible compounds, previously identified in sheep from literature, correlated to flavor attributes and sheep characteristics.

Author dc:creator

• Bechtold, Emily "Emmy" Regina

Contributors dc:contributor

• Kelly, Brendan
• Brooks, J. Chance
• Legako, Jerrad F.
• Woerner, Dale R.

• Lamb Flavor
• Rapid Evaporative Ionization Mass Spectrometry
• REIMS
• Metabolomics

Statement dc:rights

• Access is not restricted.

Language dc:language

eng

Handle dc:identifier.uri

https://hdl.handle.net/2346/87473

OAI identifier oai:identifier

oai:tdl-ir.tdl.org:2346/87473