Profiling of contemporary beer styles using liquid chromatography quadrupole time-of-flight mass spectrometry, multivariate analysis, and machine learning techniques.

Although all beer is brewed using the same four classes of ingredients, contemporary beer styles show wide variation in flavor and color, suggesting differences in their chemical profiles. A selection of 32 beers covering five styles (India pale ale, blonde, stout, wheat, and sour) were investigated to determine chemical features, which discriminate between popular beer styles. The beers were analyzed in an untargeted fashion using liquid chromatography quadrupole time-of-flight mass spectrometry (LC-QTOF-MS). The separation and detection method were tuned to include compounds from important beer components, namely iso-α-acids and phenolic compounds. Due to the sheer number of unknown compounds in beer, multivariate analysis and machine learning techniques were used to pinpoint some of the compounds most influential in distinguishing beer styles. It was determined that while many phenols and iso-α-acids were present in the beers, they were not the compounds most responsible for the variations between styles. However, it was possible to discriminate each beer style using multivariate analysis. Principal component analysis (PCA) was able to separate and cluster the individual beer samples by style. A combination of statistical tools were used to predict formulas for some of the most influential metabolites from each style. Machine learning models accurately classified patterns in the five beer styles, indicating that they can be precisely distinguished by their nonvolatile chemical profile.