AOAC 133rd Annual Meeting - Final Program

Scientific Sessions | Tuesday

SYMPOSIUM: Alternative Models for Characterizing Data 3:00 PM – 4:30 PM Plaza F Chair: Sidney Sudberg, Alkemists Laboratories

3:45 PM Variability of Cocoa Metabolites Due to Cacao Genetics, Growing Conditions and Post-Harvest Processing Helene Hopfer, Allison Brown, Patrick Dolan, Aaron Wiedemer, Pennsylvania State University The USA is home to over 200 small-scale craft chocolate makers that are emphasizing the geographical origin of the cacao beans, producing so called single origin chocolates. The assump- tion is that beans from one place have a unique flavor, however, the observed flavor differences cannot be solely attributed to a geographical location, but are rather a combination of different plant genetics, growing conditions, and postharvest processes. Using data from an on-going experiment I will show how genetics, location, and harvest time affect flavor metabolites and human flavor perception. Using genetically characterized plant material from different locations and harvests, unroasted cacao beans and roasted cocoa liquor were characterized for volatile, polyphenol, total fat, and fatty acid composition by headspace solid-phase microextraction-gas chromatography-mass spectrometry (HS-SPME-GC-MS), reversed-phase liquid chromatography with diode array detection (HPLC-DAD), time-domain nuclear magnetic resonance (TD-NMR), and GC-FID. Taken together, it was found that several classes of secondary metabolites are highly variable and differ due to plant genetics, growing location, and processing, thus, raising questions about the use of metabolite ratios as markers of geographical origin. 4:10 PM HRMS Authenticity Modeling of Mango Puree for Adulteration Dorothy Yang, Karen Yannell, James Pyke, Agilent Technologies, Inc. Food authenticity is an increasing problem amongst food manu- factures and consumers as premium ingredients are becoming more expensive and therefore more frequently adulterated. Until now, routine analysis using mass spectrometry has been limited due to the lack of user-friendly workflows. A proof of concept study using mango purees demonstrated high-throughput routine classification to detect adulterated food samples using high-res- olution mass spectrometry and simplified software tools. Three varieties of mangos were profiled using a LC/Q-TOF. Features in the data were extracted, statistics applied, and differentiat- ing features were used in a multivariate statistics model. Using MassHunter Classifier 1.0, a routine analysis tool for multivari- ate statistics models, freshly acquired samples (testing set) were applied to the model to determine if the samples were pure or adulterated. This data showed 100% accuracy when the model was first build and 2 weeks after ( n= 62). Experimental design, statistical analysis, model longevity, and software enhancements for routine analysis will be discussed.

3:05 PM Update on the Systematic Review and Meta- Analysis of Quantitative Chemical Collaborative Studies Published in JAOAC INT. from 2000 – 2019, Based on the Logarithmic Metamer pC. Sidney Sudberg, Alkemists Laboratories, Robert LaBudde, Least Cost Solutions Subject to inclusion and exclusion criteria, all collaborative stud- ies for a quantitative chemical analyte published in JAOAC INT from 2000 to 2019 were identified and the reported statistical results tabulated. The reported RSD(r) and RSD(R) values were used to estimate SD(r) and SD(L) for the logarithmic metamer pC. A statistical meta-analysis of the results was performed, and an evaluation made of possible performance requirements based on tolerance limits of the observed population of studies and compared to the HorRat system. The review is still incomplete, and this presentation is an update to the previous several years’ talks with additional studies included. New analysis is provided, and conclusions are refined. 3:25 PM Rumpelstilzchen Statistics: Spinning Proficiency Test Study ‘Straw’ into Collaborative Study ‘Gold’ Paul Wehling, Medallion Laboratories / General Mills, Robert LaBudde, Least Cost Solutions Most proficiency test (PT) studies are flawed by lack of detail about the exact methods used by participants and by lack of replication or failure to report it. The principal advantage of PT studies is a typical large number of participants available at low cost. Using a PT study dataset provided by Emerald Scientific together with a reasonable repeatability assumption, we show that, with careful analysis, it may be possible to extract useful information about reproducibility error and method equivalence, avoiding the cost of a full controlled collaborative study. 3:45 PM Why Confidence Intervals? Sidney Sudberg, Alkemists Laboratories, Robert LaBudde, Least Cost Solutions The Confidence Interval (CI), the best estimate of Measurement Uncertainty (MU) is a not well known or understood statistic, by many. In this presentation we will discuss the relationship of the True Value (TV) to the Point Estimate (PE) and the CI. We will also discuss how to calculate the CI and how to use it to quantify MU for all estimates in the laboratory.

20 SEPTEMBER 6–12, 2019 SHERATON DENVER DOWNTOWN HOTEL