Table 1.
Oligonucleotides for biomarker gene detection and sequencing.
Figure 1.
Effect of spectrum processing parameters on peak reproducibility.
Number and proportion of reproducible peaks in TY-2482 formic acid extraction replicate spectra as a function of spectrum processing parameters (A). Half window sizes for SNIP baseline correction and signal to noise ratio thresholds for peak detection are represented by symbol and fill colour, respectively. For each combination, 16 variants representing different half window sizes for smoothing (2, 4, 8, 12) and peak detection (4, 8, 12, 16) are shown. Dashed lines mark the parameter combination employed for all subsequent analyses. Representative spectra from extreme positions of the parameter space (arrows) are shown in detail (B).
Figure 2.
MALDI-TOF mass spectrum of E. coli outbreak isolate TY-2482.
Representative whole cell MALDI-TOF mass spectrum of the Shiga-Toxigenic E. coli outbreak isolate TY-2482 acquired after formic acid extraction. Inlays show enlarged views of outbreak strain specific marker peaks and the amino acid sequence of the corresponding proteins. Peptides identified by LC-MS/MS are indicated by a gray background. The tick mark interval in the enlarged peak views is set to 100.
Figure 3.
Marker peak characteristics in spectra from study isolates.
Mean signal to noise ratio of outbreak strain marker peaks (A, B) and mean signal intensitiy (C, D) at marker peak position in formic acid extraction (A, C) and direct sample deposition (B, D) triplicate spectra from 293 study isolates. Black triangles and white circles represent measurements from 104 outbreak and 189 non outbreak E. coli isolates, respectively. Red colour indicates misidentified isolates.
Table 2.
Correct classification rates (%) of MALDI-TOF typing.
Table 3.
Peak detection rates (%) among study isolates.
Table 4.
Distribution (%) of peak detection rates in triplicate spectra of outbreak isolates.
Table 5.
Variants of marker peak protein genes found among non outbreak study isolates by PCR and sequencing.
Table 6.
Isolate classification with whole spectrum similarity measures.
Figure 4.
Performance of isolate classification by whole spectrum similarity to reference spectra.
Accuracy, sensitivity and specificity for the classification of study isolates by Jaccard’s distance to TY-2482 reference spectra as a function of the selected threshold. Grey areas represent bootstrap estimates of 95% confidence intervals for thresholds derived from the distribution of distance values among outbreak isolate triplicate spectra.
Figure 5.
Whole spectrum similarity among non outbreak study isolates.
Distribution of Jaccard distance values from pairwise spectrum comparisons among non outbreak study isolates (dark grey, n = 17955) and single isolate replicate spectra (light grey, n = 189). The dashed line represents a threshold for spectral identity derived from the replicate spectrum distribution (mean+2×SD). The dotted line represents a less conservative threshold that would correctly classify 95% of all isolate pairs that were found spectrally identical upon manual spectrum comparison.