Autor(es):
Duarte, Ricardo Franco 
; Mendes, Inês ; Umek, Lan ; Neves, J. Drumonde ; Zupan, Blaz ; Schuller, Dorit
Data: 2012
Identificador Persistente: http://hdl.handle.net/1822/22571
Origem: RepositóriUM - Universidade do Minho
Assunto(s): Computational model; Yeast; Saccharomyces; Winemaking; Microsatellite; Phenotype; Bayesian classifier
Descrição
Genome sequencing is essential to understand individual variation and to study the relationship between genotype and phenotype. Recently, large-scale sequencing projects of Saccharomyces cerevisiae revealed the existence of a few well defined lineages and some mosaics of that lineages, and suggested the occurrence of two domestication events during the history of association to human activities, one for sake strains and one for wine yeasts. Although the diversity of S. cerevisiae strains in winemaking environments is rather high, suggesting the occurrence of specific natural strains associated with particular terroirs, scarce information is available regarding phenotypic variability among strains used for different biotechnological applications.
The objective of the present work was to undertake high-throughput approaches for a genetic evaluation of 172 S. cerevisiae strains from different geographical origins and technological uses (winemaking, brewing, bakery, distillery, laboratory, natural, etc.) and computationally relate the results with 30 phenotypic tests that were previously obtained. Genetic characterization was performed using eleven polymorphic S. cerevisiae specific microsatellite loci. More than 200 different alleles were obtained, being around 30 responsible for the highest strain variability. 8944 data points were generated and Principal Component Analysis (PCA) revealed the microsatellites C4 and ScYOR267c, as well as the alleles AAT5-256 and AAT6-256 as the most contributing to intra-strain variability. Based on microsatellite allelic information, the software Orange [1] was used to find sub-groups of strains with similar values of phenotypic features and microsatellite allelic patterns. Globally, our study demonstrates that computational approaches can be successfully used to estimate a strain’s biotechnological potential from genotypic data, simplifying laborious strain selection programs by partially replacing phenotypic screens through a preliminary selection based on a strain’s microsatellite allelic combinations.
