Genome-based model for differentiating between infection and carriage Staphylococcus aureus

We address the disease-associated k-mers by using a comprehensive genome-wide association study (GWAS) to compare genetic variation of <i>S. aureus </i>isolates. Due to high-dimensional genomic data, a two-stage analysis process was performed to identify the disease-associated k-mers by multiple GWAS methods including the Linear Mixed Model (LMM),the phylogeny-based method (Scoary), the regularized regression model (Least Absolute Shrinkage and Selection Operator, LASSO) and the machine learning method (Random Forest, RF). The importance of each k-mer predictor was sorted by the mean decrease in impurity (Mean Decrease Gini, MDG).