LASSO and SVM: an alternative approach to identify associated genome regions for simple and complex traits in cattle

Variable selection via the LASSO regression analysis followed by a classification of phenotypes into two groups of cases and controls using SVM was performed to identify genomic regions associated with phenotypic traits of interest. This approach was successfully validated by identifying previously known genomic associations for two Mendelian traits and a complex polygenic trait in Swiss cattle populations. Furthermore, new associated regions for the trait multiple birth were found in Holstein. Therefore, this approach proved to be a valuable alternative to identify associated genomic regions for inherited traits in livestock.