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.

A SARS-CoV-2 sequence submission tool for the European Nucleotide Archive.

SUMMARY: Many aspects of the global response to the COVID-19 pandemic are enabled by the fast and open publication of SARS-CoV-2 genetic sequence data. The European Nucleotide Archive (ENA) is the European recommended open repository for genetic sequences. In this work, we present a tool for submitting raw sequencing reads of SARS-CoV-2 to ENA. The tool features a single-step submission process, a graphical user interface, tabular-formatted metadata and the possibility to remove human reads prior to submission. A Galaxy wrap of the tool allows users with little or no bioinformatics knowledge to do bulk sequencing read submissions. The tool is also packed in a Docker container to ease deployment. AVAILABILITY AND IMPLEMENTATION: CLI ENA upload tool is available at github.com/usegalaxy-eu/ena-upload-cli (DOI 10.5281/zenodo.4537621); Galaxy ENA upload tool at toolshed.g2.bx.psu.edu/view/iuc/ena_upload/382518f24d6d and github.com/galaxyproject/tools-iuc/tree/master/tools/ena_upload (development); and ENA upload Galaxy container at github.com/ELIXIR-Belgium/ena-upload-container (DOI 10.5281/zenodo.4730785).

Dissection of the molecular bases of genotype x environment interactions: a study of phenotypic plasticity of Saccharomyces cerevisiae in grape juices.

BACKGROUND: The ability of a genotype to produce different phenotypes according to its surrounding environment is known as phenotypic plasticity. Within different individuals of the same species, phenotypic plasticity can vary greatly. This contrasting response is caused by gene-by-environment interactions (GxE). Understanding GxE interactions is particularly important in agronomy, since selected breeds and varieties may have divergent phenotypes according to their growing environment. Industrial microbes such as Saccharomyces cerevisiae are also faced with a large range of fermentation conditions that affect their technological properties. Finding the molecular determinism of such variations is a critical task for better understanding the genetic bases of phenotypic plasticity and can also be helpful in order to improve breeding methods. RESULTS: In this study we implemented a QTL mapping program using two independent cross (~ 100 progeny) in order to investigate the molecular basis of yeast phenotypic response in a wine fermentation context. Thanks to whole genome sequencing approaches, both crosses were genotyped, providing saturated genetic maps of thousands of markers. Linkage analyses allowed the detection of 78 QTLs including 21 with significant interaction with the environmental conditions. Molecular dissection of a major QTL demonstrated that the sulfite pump Ssu1p has a pleiotropic effect and impacts the phenotypic plasticity of several traits. CONCLUSIONS: The detection of QTLs and their interactions with environment emphasizes the complexity of yeast industrial traits. The validation of the interaction of SSU1 allelic variants with the nature of the fermented juice increases knowledge about the impact of the sulfite pump during fermentation. All together these results pave the way for exploiting and deciphering the genetic determinism of phenotypic plasticity.

Transition between fermentation and respiration determines history-dependent behavior in fluctuating carbon sources.

Cells constantly adapt to environmental fluctuations. These physiological changes require time and therefore cause a lag phase during which the cells do not function optimally. Interestingly, past exposure to an environmental condition can shorten the time needed to adapt when the condition re-occurs, even in daughter cells that never directly encountered the initial condition. Here, we use the molecular toolbox of Saccharomyces cerevisiae to systematically unravel the molecular mechanism underlying such history-dependent behavior in transitions between glucose and maltose. In contrast to previous hypotheses, the behavior does not depend on persistence of proteins involved in metabolism of a specific sugar. Instead, presence of glucose induces a gradual decline in the cells' ability to activate respiration, which is needed to metabolize alternative carbon sources. These results reveal how trans-generational transitions in central carbon metabolism generate history-dependent behavior in yeast, and provide a mechanistic framework for similar phenomena in other cell types.

High-throughput system-wide engineering and screening for microbial biotechnology.

Genetic engineering and screening of large number of cells or populations is a crucial bottleneck in today's systems biology and applied (micro)biology. Instead of using standard methods in bottles, flasks or 96-well plates, scientists are increasingly relying on high-throughput strategies that miniaturize their experiments to the nanoliter and picoliter scale and the single-cell level. In this review, we summarize different high-throughput system-wide genome engineering and screening strategies for microbes. More specifically, we will emphasize the use of multiplex automated genome evolution (MAGE) and CRISPR/Cas systems for high-throughput genome engineering and the application of (lab-on-chip) nanoreactors for high-throughput single-cell or population screening.

The yeast TUM1 affects production of hydrogen sulfide from cysteine treatment during fermentation.

The undesirable rotten-egg odour of hydrogen sulfide (H2S) produced by yeast shortly after yeast inoculation of grape musts might be an important source of desirable varietal thiols, which contribute to tropical aromas in varieties such as Sauvign-on Blanc. In this study, we observed that Saccharomyces cerevisiae strains produce an early burst of H2S from cysteine. Both Δmet2 and Δmet17 strains produce a larger burst, likely because they are unable to utilise the H2S in the sulfate assimilation pathway. For the first time, we show that TUM1 is partly responsible for the early production of H2S from cysteine. Overex-pressing TUM1 elevated production of H2S, whilst its deletion yields only half of the H2S. We further confirmed that yeast convert cysteine to H2S by analysing growth of mutants lacking components of the transsulfuration pathway. High concent-rations of cysteine overcame this growth block, but required TUM1 Collectively, the data indicate that S. cerevisiae does not convert cysteine to sulfate or sulfite, but rather to sulfide via a novel pathway that requires the action of Tum1p. The findi-ngs of this study may allow the improvement of commercial yeasts through the manipulation of sulfur metabolism that are better suited towards production of fruit-driven styles.

Domestication and Divergence of Saccharomyces cerevisiae Beer Yeasts.

Whereas domestication of livestock, pets, and crops is well documented, it is still unclear to what extent microbes associated with the production of food have also undergone human selection and where the plethora of industrial strains originates from. Here, we present the genomes and phenomes of 157 industrial Saccharomyces cerevisiae yeasts. Our analyses reveal that today's industrial yeasts can be divided into five sublineages that are genetically and phenotypically separated from wild strains and originate from only a few ancestors through complex patterns of domestication and local divergence. Large-scale phenotyping and genome analysis further show strong industry-specific selection for stress tolerance, sugar utilization, and flavor production, while the sexual cycle and other phenotypes related to survival in nature show decay, particularly in beer yeasts. Together, these results shed light on the origins, evolutionary history, and phenotypic diversity of industrial yeasts and provide a resource for further selection of superior strains. PAPERCLIP.

Genetic characterization of strains of Saccharomyces uvarum from New Zealand wineries.

We present a genetic characterization of 65 isolates of Saccharomyces uvarum isolated from wineries in New Zealand, along with the complete nucleotide sequence of a single sulfite-tolerant isolate. The genome of the New Zealand isolate averaged 99.85% nucleotide identity to CBS7001, the previously sequenced strain of S. uvarum. However, three genomic segments (37-87 kb) showed 10% nucleotide divergence from CBS7001 but 99% identity to Saccharomyces eubayanus. We conclude that these three segments appear to have been introgressed from that species. The nucleotide sequence of the internal transcribed spacer (ITS) region from other New Zealand isolates were also very similar to that of CBS7001, and hybrids showed complete genetic compatibility for some strains, with tetrads giving four viable progeny that showed 2:2 segregations of marker genes. Some strains showed high tolerance to sulfite, with genetic analysis indicating linkage of this trait to the transcription factor FZF1, but not to SSU1, the sulfite efflux pump that it regulates in order to confer sulfite tolerance in Saccharomyces cerevisiae. The fermentation characteristics of selected strains of S. uvarum showed exceptionally good cold fermentation characteristics, superior to the best commercially available strains of S. cerevisiae.

MET2 affects production of hydrogen sulfide during wine fermentation.

The production of hydrogen sulfide (H2S) during yeast fermentation contributes negatively to wine aroma. We have mapped naturally occurring mutations in commercial wine strains that affect production of H2S. A dominant R310G mutant allele of MET2, which encodes homoserine O-acetyltransferase, is present in several wine yeast strains as well as in the main lab strain S288c. Reciprocal hemizygosity and allele swap experiments demonstrated that the MET2 R310G allele confers reduced H2S production. Mutations were also identified in genes encoding the two subunits of sulfite reductase, MET5 and MET10, which were associated with reduced H2S production. The most severe of these, an allele of MET10, showed five additional phenotypes: reduced growth rate on sulfate, elevated secretion of sulfite, and reduced production in wine of three volatile sulfur compounds: methionol, carbon disulfide and methylthioacetate. Alleles of MET5 and MET10, but not MET2, affected H2S production measured by colour assays on BiGGY indicator agar, but MET2 effects were seen when bismuth was added to agar plates made with Sauvignon blanc grape juice. Collectively, the data are consistent with the hypothesis that H2S production during wine fermentation results predominantly from enzyme activity in the sulfur assimilation pathway. Lower H2S production results from mutations that reduce the activity of sulfite reductase, the enzyme that produces H2S, or that increase the activity of L-homoserine-O-acetyltransferase, which produces substrate for the next step in the sulfur assimilation pathway.

The yeast IRC7 gene encodes a ß-lyase responsible for production of the varietal thiol 4-mercapto-4-methylpentan-2-one in wine.

Three varietal thiols are key aroma compounds in Sauvignon Blanc wines: 4-mercapto-4-methylpentan-2-one (4MMP), 3-mercaptohexanol (3MH) and its acetylated derivative 3-mercaptohexyl acetate (3MHA). Screening of Saccharomyces cerevisiae strains identified a clinical isolate with elevated 4MMP production after fermentation. Bulked Segregant Analysis of a cross between this isolate and the laboratory strain revealed a single major locus for 4MMP production near the telomere of chromosome 6. Deletion of the IRC7 gene from this region in YJM450 reduced 4MMP production below detectable levels, but did not affect yields of 3MH, in Sauvignon Blanc wine. Sequencing revealed that the IRC7 gene in YJM450 had been introgressed from a strain of Saccharomyces paradoxus. Most strains of S. cerevisiae, including the laboratory strain S288C, have a 38-bp deletion that inactivates IRC7. Overexpression of a full-length S. cerevisiae allele of IRC7 in a wine yeast, Zymaflore F15, increased 4MMP production in Sauvignon Blanc wine from undetectable levels (<10 ng L(-1)) to concentrations of 1000 ng L(-1), and also increased 3MH and 3MHA. Biochemical analysis of soluble protein extracts showed that both the cerevisiae and paradoxus IRC7 proteins show ß-lyase activity, with a substrate preference for cys-4MMP over cys-3MH.