Predicting and improving complex beer flavor through machine learning.

The perception and appreciation of food flavor depends on many interacting chemical compounds and external factors, and therefore proves challenging to understand and predict. Here, we combine extensive chemical and sensory analyses of 250 different beers to train machine learning models that allow predicting flavor and consumer appreciation. For each beer, we measure over 200 chemical properties, perform quantitative descriptive sensory analysis with a trained tasting panel and map data from over 180,000 consumer reviews to train 10 different machine learning models. The best-performing algorithm, Gradient Boosting, yields models that significantly outperform predictions based on conventional statistics and accurately predict complex food features and consumer appreciation from chemical profiles. Model dissection allows identifying specific and unexpected compounds as drivers of beer flavor and appreciation. Adding these compounds results in variants of commercial alcoholic and non-alcoholic beers with improved consumer appreciation. Together, our study reveals how big data and machine learning uncover complex links between food chemistry, flavor and consumer perception, and lays the foundation to develop novel, tailored foods with superior flavors.

Polygenic Analysis of Tolerance to Carbon Dioxide Inhibition of Isoamyl Acetate "Banana" Flavor Production in Yeast Reveals MDS3 as Major Causative Gene.

The introduction in modern breweries of tall cylindroconical fermentors, replacing the traditional open fermentation vats, unexpectedly revealed strong inhibition of flavor production by the high CO2 pressure in the fermentors. We have screened our collection of Saccharomyces cerevisiae strains for strains displaying elevated tolerance to inhibition of flavor production by +0.65 bar CO2, using a laboratory scale CO2 pressurized fermentation system. We focused on the production of isoamyl acetate, a highly desirable flavor compound conferring fruity banana flavor in beer and other alcoholic beverages, from its precursor isoamyl alcohol (IAAc/Alc ratio). We selected the most tolerant Saccharomyces cerevisiae strain, saké yeast Kyokai no. 1, isolated a stable haploid segregant seg63 with the same high IAAc/Alc ratio under CO2 pressure, crossed seg63 with the unrelated inferior strain ER7A and phenotyped 185 haploid segregants, of which 28 displaying a high IAAc/Alc ratio were pooled. Mapping of Quantitative Trait Loci (QTLs) by whole-genome sequence analysis based on SNP variant frequency revealed two QTLs. In the major QTL, reciprocal hemizygosity analysis identified MDS3 as the causative mutant gene, a putative member of the TOR signaling pathway. The MDS3Seg.63 allele was dominant and contained a single causative point mutation, T2171C, resulting in the F274S substitution. Introduction of MDS3Seg.63 in an industrial tetraploid lager yeast with CRISPR/Cas9 enhanced isoamyl acetate production by 145% under CO2 pressure. This work shows the strong potential of polygenic analysis and targeted genetic modification for creation of cisgenic industrial brewer's yeast strains with specifically improved traits. IMPORTANCE The upscaling of fermentation to very tall cylindroconical tanks is known to negatively impact beer flavor. Most notably, the increased CO2 pressure in such tanks compromises production by the yeast of the desirable fruity "banana" flavor (isoamyl acetate). The cause of the CO2 inhibition of yeast flavor production has always remained enigmatic. Our work has brought the first insight into its molecular-genetic basis and provides a specific gene tool for yeast strain improvement. We first identified a yeast strain with superior tolerance to CO2 inhibition of flavor production, and applied polygenic analysis to identify the responsible gene. We narrowed down the causative element to a single nucleotide difference, MDS3T2171C, and showed that it can be engineered into brewing yeast to obtain strains with superior flavor production in high CO2 pressure conditions, apparently without affecting other traits relevant for beer brewing. Alternatively, such a strain could be obtained through marker-assisted breeding.

Interspecific hybridization facilitates niche adaptation in beer yeast.

Hybridization between species often leads to non-viable or infertile offspring, yet examples of evolutionarily successful interspecific hybrids have been reported in all kingdoms of life. However, many questions on the ecological circumstances and evolutionary aftermath of interspecific hybridization remain unanswered. In this study, we sequenced and phenotyped a large set of interspecific yeast hybrids isolated from brewing environments to uncover the influence of interspecific hybridization in yeast adaptation and domestication. Our analyses demonstrate that several hybrids between Saccharomyces species originated and diversified in industrial environments by combining key traits of each parental species. Furthermore, posthybridization evolution within each hybrid lineage reflects subspecialization and adaptation to specific beer styles, a process that was accompanied by extensive chimerization between subgenomes. Our results reveal how interspecific hybridization provides an important evolutionary route that allows swift adaptation to novel environments.

Brettanomyces yeasts--From spoilage organisms to valuable contributors to industrial fermentations.

Ever since the introduction of controlled fermentation processes, alcoholic fermentations and Saccharomyces cerevisiae starter cultures proved to be a match made in heaven. The ability of S. cerevisiae to produce and withstand high ethanol concentrations, its pleasant flavour profile and the absence of health-threatening toxin production are only a few of the features that make it the ideal alcoholic fermentation organism. However, in certain conditions or for certain specific fermentation processes, the physiological boundaries of this species limit its applicability. Therefore, there is currently a strong interest in non-Saccharomyces (or non-conventional) yeasts with peculiar features able to replace or accompany S. cerevisiae in specific industrial fermentations. Brettanomyces (teleomorph: Dekkera), with Brettanomyces bruxellensis as the most commonly encountered representative, is such a yeast. Whilst currently mainly considered a spoilage organism responsible for off-flavour production in wine, cider or dairy products, an increasing number of authors report that in some cases, these yeasts can add beneficial (or at least interesting) aromas that increase the flavour complexity of fermented beverages, such as specialty beers. Moreover, its intriguing physiology, with its exceptional stress tolerance and peculiar carbon- and nitrogen metabolism, holds great potential for the production of bioethanol in continuous fermentors. This review summarizes the most notable metabolic features of Brettanomyces, briefly highlights recent insights in its genetic and genomic characteristics and discusses its applications in industrial fermentation processes, such as the production of beer, wine and bioethanol.

Glucose and sucrose: hazardous fast-food for industrial yeast?

Yeast cells often encounter a mixture of different carbohydrates in industrial processes. However, glucose and sucrose are always consumed first. The presence of these sugars causes repression of gluconeogenesis, the glyoxylate cycle, respiration and the uptake of less-preferred carbohydrates. Glucose and sucrose also trigger unexpected, hormone-like effects, including the activation of cellular growth, the mobilization of storage compounds and the diminution of cellular stress resistance. In an industrial context, these effects lead to several yeast-related problems, such as slow or incomplete fermentation, 'off flavors' and poor maintenance of yeast vitality. Recent studies indicate that the use of mutants with altered responses to carbohydrates can significantly increase productivity. Alternatively, avoiding unnecessary exposure to glucose and sucrose could also improve the performance of industrial yeasts.