Fig. 6: Impact of shape, features and heterogeneity on classification accuracy. | ISME Communications

Fig. 6: Impact of shape, features and heterogeneity on classification accuracy.

From: Fast quantification of gut bacterial species in cocultures using flow cytometry and supervised classification

Fig. 6

A F1 score for pairwise in silico predictions within 12 species (66 combinations) grouped by shape (i.e., Bacillus, coccus, bifid, coccobacillus). B Feature importance calculated with Lime. C–E Heterogeneity assessed for seven biological replicates, for four species, Bact. thetaiotaomicron (BT), Bl. hydrogenotrophica (BH), R. intestinalis (RI), E. coli (EC), and medium debris (Blank). C Intra-cluster variation is computed as the mean pairwise Euclidean distance averaged across experiments per species. D Inter-cluster variation is assessed as the mean of all pairwise centroid distances per species, where a centroid is computed for each species-specific experiment. Accuracy refers to CellScanner accuracy when assigning events from merged replicates to the correct experiment of origin. A low intra-cluster variation combined with a high inter-cluster variation reduces the overlap between experiment-specific clusters and results in a high accuracy.

Back to article page