Figures & data
Table 1. Average results in accuracy terms by repeating 20 times a 10-fold cross validation analysis 20 times for all ML supervised algorithms for each method and step.
Figure 1. Distance plot of RF and PLS-DA for discrimination between ESRI developers susceptible to P/T and their isogenic isolates resistant to P/T. A. C, Full Spectrum method. B. D, Threshold method. ESRI: Extended spectrum resistance to BL/BLI susceptible to P/T; Pressed: ESRI isogenic isolates resistant to P/T after pressure with P/T; PLS-DA: partial least squares discriminant analysis; RF: random forest. t0 and t1 are two components of an algorithm called MDS that help to represent data from a distance matrix.
![Figure 1. Distance plot of RF and PLS-DA for discrimination between ESRI developers susceptible to P/T and their isogenic isolates resistant to P/T. A. C, Full Spectrum method. B. D, Threshold method. ESRI: Extended spectrum resistance to BL/BLI susceptible to P/T; Pressed: ESRI isogenic isolates resistant to P/T after pressure with P/T; PLS-DA: partial least squares discriminant analysis; RF: random forest. t0 and t1 are two components of an algorithm called MDS that help to represent data from a distance matrix.](/cms/asset/5ca0a788-aa1c-41de-a190-dd5811e56bad/temi_a_2113746_f0001_oc.jpg)
Figure 2. Distance plot of RF and PLS-DA for discrimination between P/T-susceptible and P/T-resistant E. coli (first step analysis). A. C, Full Spectrum method. B. D, Threshold method. PLS-DA: Partial Least Square Discriminant Analysis; RF: Random Forest. t0 and t1 are two components of an algorithm called MDS that help to represent data from a distance matrix.
![Figure 2. Distance plot of RF and PLS-DA for discrimination between P/T-susceptible and P/T-resistant E. coli (first step analysis). A. C, Full Spectrum method. B. D, Threshold method. PLS-DA: Partial Least Square Discriminant Analysis; RF: Random Forest. t0 and t1 are two components of an algorithm called MDS that help to represent data from a distance matrix.](/cms/asset/85e328a6-f23b-4f05-a94e-8d4a8e8ea624/temi_a_2113746_f0002_oc.jpg)
Figure 3. Distance plot of RF and PLS-DA for discrimination between P/T-susceptible and ESRI developer Escherichia coli (second step analysis). A. C, Full Spectrum method. B. D, Threshold method. PLS-DA: Partial Least Square Discriminant Analysis; RF: Random Forest. t0 and t1 are two components of an algorithm called MDS that help to represent data from a distance matrix.
![Figure 3. Distance plot of RF and PLS-DA for discrimination between P/T-susceptible and ESRI developer Escherichia coli (second step analysis). A. C, Full Spectrum method. B. D, Threshold method. PLS-DA: Partial Least Square Discriminant Analysis; RF: Random Forest. t0 and t1 are two components of an algorithm called MDS that help to represent data from a distance matrix.](/cms/asset/64bf6503-bed0-42b4-bede-272cf239ac71/temi_a_2113746_f0003_oc.jpg)
Figure 4. Feature importance from RF analysis from each assay. A, ESRI developers susceptible to P/T and their isogenic isolates resistant to P/T. B, P/T-susceptible and P/T-resistant E. coli (Step 1). C, P/T-susceptible and ESRI developers susceptible to P/T (Step 2). RF: Random Forest; ESRI: Extended Spectrum Resistance to BL/BLI susceptible to P/T.
![Figure 4. Feature importance from RF analysis from each assay. A, ESRI developers susceptible to P/T and their isogenic isolates resistant to P/T. B, P/T-susceptible and P/T-resistant E. coli (Step 1). C, P/T-susceptible and ESRI developers susceptible to P/T (Step 2). RF: Random Forest; ESRI: Extended Spectrum Resistance to BL/BLI susceptible to P/T.](/cms/asset/35658458-3b29-4b28-959c-dff9c965e850/temi_a_2113746_f0004_oc.jpg)
Figure 5. Average ROC curves from first 10-folds and their AUC values obtained from RF using Full Spectrum method to discriminate between A, ESRI developers susceptible to P/T and their isogenic isolates resistant to P/T B, P/T-susceptible and P/T-resistant E. coli and C, P/T-susceptible and ESRI developer Escherichia coli. Isogenic isolates resistant to P/T, resistant isolates to P/T and ESRI developer isolates susceptible to P/T were labelled as positive categories respectively.
![Figure 5. Average ROC curves from first 10-folds and their AUC values obtained from RF using Full Spectrum method to discriminate between A, ESRI developers susceptible to P/T and their isogenic isolates resistant to P/T B, P/T-susceptible and P/T-resistant E. coli and C, P/T-susceptible and ESRI developer Escherichia coli. Isogenic isolates resistant to P/T, resistant isolates to P/T and ESRI developer isolates susceptible to P/T were labelled as positive categories respectively.](/cms/asset/ec12869f-1a7c-48fc-937a-a4bd6309efc6/temi_a_2113746_f0005_oc.jpg)
Supplemental Material
Download MS Word (30.4 KB)Data availability statement
The data that support the findings of this study are available from the corresponding author upon reasonable request.