Fig 1.
Reproducibility of MALDI-TOF MS based typing data.
Box-plots show the average similarities in percentage. Median and interquartile range is indicated, whiskers indicate the range. Numbers indicate the different centers, (e.g. center 1 etc.); center 5A and 5B indicates that two microbiologists independently performed the experiment at center 5. (A) Technical reproducibility in percent. (B) Biological reproducibility in percent. (C) Peak comparison between centers. Three peaks are shown: peaks at 9000 and 9220 m/z are detected by all centers, however the peak at 9060 is only detected by one center.
Fig 2.
Principal component and discriminant analysis.
(A) Principal component analysis (PCA) indicates a strong center specific effect. Each center is depicted with a different colour code, e.g. red data points clusters together and do not allow to separate clusters independent of each center. (B) PCA within the groups of all clusters implicates some difference, however the clusters cannot reliably be separated. (C) The discriminant analysis highlights that the clusters can be separated. Each cluster is shown with a different colour code.
Fig 3.
Identification of cluster-separating peaks. The x-axis indicates the m/z value and the y-axis the peak intensity. Alternative 1: Single line analysis shows an extract of three peak profiles with a ‘peak pair’ at 9712 and 9739 m/z. The peak on the left part of the graph shows a shift in the top two spectra compared to the bottom spectrum, while the peak on the right is in the same position for all three. Alternative 2: The overlay mode of multiple spectra show ‘peak pairs’ at 9712 and 9739 m/z. Each line reflects a single measurement. The green lines are spectra of cluster 2-isolates.
Table 1.
Peaks identified to separate the clusters.
Fig 4.
(A) Separating capacity for reference center: average scores (p-values) and standard deviation with the own group of the spectra (green) and with the best scoring other group (red). For each group the score with the own group is highest with no overlap in the standard deviation of both scores. (B) Separating capacity for all centers: the number of isolates belonging to a certain predicted group by SVM per group they actually belong to. The diagonal contains the number of correct identifications per group, outside the diagonal are the incorrect identifications.