Figure 1

Illustration of the pan-genome approach for identifying candidate virulence-associated genes (VAGs). A The first data set analyzed consisted of all 208 isolates and B the second data set consisted of a subset of 161 isolates. Genes with a significant association (p-value < 0.05) with pathotype, and present in more than 50% (69/139) of isolates within the pathogenic pathotype and present in less than 50% (11/22) of isolates within the commensal pathotype, were analyzed using the LASSO shrinkage regression model. Genes identified in both LASSO analyses were selected as the ‘best’ predictors of pathotype.