Optimal settings of fingerprint-type analysing computer software for the analysis of enterohaemorrhagic Escherichia coli pulsed-field gel electrophoresis patterns

Abstract

Settings of fingerprint-type analysing computer software were optimized for analysis of enterohaemorrhagic Escherichia coli (EHEC) pulsed-field gel electrophoresis (PFGE) patterns. Under the lowest values of parameters, maximum value of similarities calculated using the Dice coefficient were obtained between PFGE patterns from one EHEC strain on the same gel when reference lanes for calibration of distortions during electrophoresis were set to every fourth lane. PFGE patterns of 15 EHEC strains on different gels were investigated. Similarity values calculated using the Pearson product-moment correlation coefficient (Pearson correlation) were significantly higher than those using the Dice coefficient with optimal values of parameters determined by the program (P<0·01). When PFGE patterns of 45 EHEC strains were analysed by the computer program, EHEC strains from one mass outbreak and three intra-family outbreaks were each clustered and the similarity values within the clusters were >90% using Pearson correlation.