Fig. 5From: Comparison of 16S and whole genome dog microbiomes using machine learningThe ROC curve of the random forest model’s performance and most important features used in classifying weight phenotype for WGS (A, B) and 16S (C, D). The importance score for a feature is the fraction of samples across all trees that traverse a node and split on that feature. Notably, B. coprocola was the most important feature for classifying the weight phenotype in WGS samplesBack to article page