OBJECTIVES: Methicillin-resistant Staphylococcus aureus (MRSA) causes hospital- and community-acquired infections. It is not clear whether genetic characteristics of the bacteria contribute to disease pathogenesis in MRSA infection. We hypothesized that whole genome analysis of MRSA strains could reveal the key gene loci and/or the gene mutations that affect clinical manifestations of MRSA infection. METHODS: Whole genome sequences (WGS) of MRSA of 154 strains were analyzed with respect to clinical manifestations and data. Further, we evaluated the association between clinical manifestations in MRSA infection and genomic information. RESULTS: WGS revealed gene mutations that correlated with clinical manifestations of MRSA infection. Moreover, 12 mutations were selected as important mutations by Random Forest analysis. Cluster analysis revealed strains associated with a high frequency of bloodstream infection (BSI). Twenty seven out of 34 strains in this cluster caused BSI. These strains were all positive for collagen adhesion gene (cna) and have mutations in the locus, those were selected by Random Forest analysis. Univariate and multivariate analysis revealed that these gene mutations were the predictor for the incidence of BSI. Interestingly, mutant CNA protein showed lower attachment ability to collagen, suggesting that the mutant protein might contribute to the dissemination of bacteria. CONCLUSIONS: These findings suggest that the bacterial genotype affects the clinical characteristics of MRSA infection.