Sensory-classified roasted arabica coffee with GC×GC-TOFMS and chemometrics (Caitlin Cain, MDCW 2024)
- Photo: MDCW: Sensory-classified roasted arabica coffee with GC×GC-TOFMS and chemometrics (Caitlin Cain, MDCW 2024)
- Video: LabRulez: Caitlin Cain: Sensory-classified roasted arabica coffee with GC×GC-TOFMS and chemometrics (MDCW 2024)
🎤 Presenter: Caitlin Cain¹, Meriem Gaida², Pierre-Hugues Stefanuto², Jean-François Focant², Robert Synovec¹, Susan Jackels³, Kristen Skogerboe³ (¹University of Washington, Seattle, USA, ²University of Liège, Liège, Belgium, ³Seattle University, Seattle, USA)
💡 Book in your calendar: 16th Multidimensional Chromatography Workshop (MDCW) 3 - 5. February 2025
Abstract
As coffee consumption increases globally, demands for specialty and/or high-quality coffee beans are primarily driven by consumer preferences toward the taste and aroma of the final brew. However, the presence of chemical defects can negatively impact the sensory properties of coffee. Potato taste defect (PTD), unique to East African coffee, is characterized by a distinct musty, vegetable, raw potato aroma due to the presence of 2-isopropyl-3-methoxypyrazine (IPMP). This work develops a volatile fingerprint of PTD in roasted arabica coffee using headspace-solid-phase microextraction-comprehensive two-dimensional gas chromatography with time-of-flight mass spectrometry (HS-SPME-GC×GC-TOFMS) and chemometrics.
Tile-based Fisher ratio (F-ratio) analysis discovered over 300 volatiles that discriminated clean (i.e., no off-odor) coffee samples from those identified as having a strong PTD odor (p-value less than 0.01). Analytes associated with unpleasant sensory descriptions were positively correlated with the odor of PTD, while analytes associated with desirable aromas were found to be negatively correlated with odor severity. Principal components analysis (PCA) highlighted that the coffee samples clustered based on the presence of PTD. Furthermore, use of the class-distinguishing analytes discovered by F-ratio analysis for partial least squares (PLS) regression produced accurate models of IPMP concentration with low prediction errors (~ 10 % error). Investigation of the PCA and PLS loadings revealed the volatile compounds which are most important to predicting PTD in roasted coffee. Ultimately, these results reveal the complex volatile signature of PTD in East African coffee beans, demonstrating that the defect has outcomes that extend well beyond the concentration of IPMP.
