Fig. 1
Identified discriminators for differentiation between Gram-negative and Gram-positive bacteria. Boxplot of random forest method identified discriminators for the training (left) and validation (right) set for differentiation into Gram-negative (white boxplots) and Gram-positive (grey boxplots) bacteria in anaerobic samples. Discriminators are given at their mass to charge ratio (m/z) of appearance. H2 (m/z = 2) was identified using electron impact ionization. Compounds detected at m/z = 34–36, 64 and 66 were measured by chemical ionization using mercury as primary ion. Compounds detected at m/z = 35, 64, 76 and 80 were measured by chemical ionization using xenon as primary ion. Signal intensity is given in counts per second (cps). p values were computed using a Kolmogorov–Smirnov test for testing if distributions were equal in both groups. Due to a previous H2 calibration, negative H2 values were obtained. Therefore, for the graphical presentation a uniform projection of the H2 values into the positive was performed