Extended Data Fig. 2: Compound induced phenotypic profile similarity derived with random forest analysis.
From: A novel antibiotic class targeting the lipopolysaccharide transporter
For the random forest model generation, analysis was applied to the reference compound set Doxycycline (Red), Levofloxacin (Green), Mecillinam (Pale blue), Meropenem (Orange), Colistin (Dark blue), RO7202110 (Magenta) as well as samples with no compound (DMSO, black) as previously described in Zoffmann et al with data from three independent experiments run in triplicates. a, In the left panel is a similarity-based projection of the correlating random forest distance matrix into 3 dimensions showing clear separation between data points belonging to different compounds and close proximity of those belonging to the same compounds, forming clusters. For the random forest analysis data proximity analysis from 4x and 8x LOED are pooled and the number of datapoints are reduced by down sampling to be equal for all reference conditions the maximal projection of the analysis. b, In the right panel an out-of-bag validation of the random forest classification model is indicated for the reference compounds. For the test compounds (MCPs RG7075573 and zosurabalpin (RG6006, ZAB), polymyxin B and the two LpxC inhibitors) a similarity score was derived from the model. The degree of frequency of matching prediction correspond to the length of the individual bar segment, where a longer segment represents a higher similarity. Thus there is a high similarity for polymyxin with >70% similarity score towards colistin fitting with the known similar MOA, while the LpxC inhibitors as expected does not have a strong correlation with any of the 6 reference compounds all known to have different MOA. For the macrocycles RG7075573 and zosurabalpin (RG6006, ZAB) > 70% similarity was seen towards the reference compound RO7202110 supporting a similar MOA for the three.