Computer-aided visual matching of derivative plots shows excellent performance Since the performance of proposed automated identification approach followed by matching the peaks positions has not reached the accuracy of identification based on traditional RAPD fingerprints, we further looked for other ways to best interpret the information present in melting curves. Simple visual inspection of a derivative curve obtained with the examined strain and its comparison to sets of curves obtained with isolates belonging to each clearly delineated species

genotype appeared intuitively as the most promising alternative. To achieve this comparison Selleck Blasticidin S in an easy-to-manage way we developed a simple computer-aided plotting scheme. Using Microsoft Excel 2007 software, plots of all derivative curves assigned to each species/genotype were prepared in separate sheets using thin lines and the curve of a tested isolate was then imported into another sheet and automatically plotted into each of the plots using a bold line. Then, all of the plots of specific species/genotypes including

the bold curve of the tested isolate were inspected visually and the best match was evaluated based on subjective judgment (see Figure 16 for an example). This evaluation was performed independently by two people in a blinded fashion, i.e. the evaluating person did not know the identity of any of the tested curves and the curves were selected in a random order for evaluation to avoid any bias. Later, a third person evaluated the accuracy of this subjective visual identification using selleck chemical a key generated during randomization. Altogether, 316 and 317 of 322 isolates were identified correctly, achieving excellent accuracy of 98.1-98.4% (for results in individual species see Table 2). In other words, 6 Lck AZD5363 concentration strains were misidentified by one evaluator and 5 strains by the other, where the 6 strains misidentified by one evaluator included the 5 strains misidentified by the other. This

concordance indicates clearly that this failure was not caused by subjective error, but rather by lack of typical properties in the misidentified melting profiles. Closer inspection of the misidentified strains showed that they included one strain which showed a completely unique fingerprint and therefore was not identified by traditional RAPD fingerprinting, and other 2 strains which showed less characteristic fingerprints, albeit it was possible to identify them using traditional RAPD fingerprinting. Figure 16 Visual matching of derivative curves as used for species identification. Plots of derivative curves obtained with all strains assigned to 9 selected species/genotypes versus the derivative curve obtained with a tested isolate are shown as an example to illustrate the visual matching approach.