Figure 5.

3D Principal component analysis of antibiotic-treated E. coli 25922 using unweighted variables from cytological imaging with associated agglomerative hierarchical clustering (AHC) and heat map made in ClustVis⁴⁷. PCA demonstrates ambiguous association with any antibiotic class but AHC demonstrates close association with both membrane active compounds and DNA synthesis inhibitors. Three independent cultures of bacteria were treated with each antibiotic and the corresponding segmentation metrics were averaged. PC1 (51.95%) vs PC2 (18.11%) vs PC3 (11.87%). 3D scatterplot was used for E. coli and not B. subtilis because only two principal components insufficiently describes variation in E. coli morphology based on cumulative variability. Two components were sufficient to capture >90% of cumulative variability in B. subtilis. DPA 154 appears to be more closely related to the action of polymyxin-B than ciprofloxacin, but clear similarity between both is demonstrated in the clustering heat map. Summary of variables that contribute to each principal component/factor, parameters for cytological profiling, and bacterial measurements may be found in the supplemental figures (Figs S1 and S2, Tables S7, S12 and 13).