MegaR: an interactive R package for rapid sample classification and phenotype prediction using metagenome profiles and machine learning

Table 2 Accuracy of optimized RF and cross validation results across three datasets from DIABIMMUNE research group

Dataset	Data type	Optimal model parameter	Model accuracy	95% CI	Cross validation accuracy
Three country cohort	16S	80%, 100T, 20P	0.9028	0.9382–0.8562	0.8685
Three country cohort	WGS	70%, 100T, 10P	0.8864	0.8312–0.9285	0.8803
T1D cohort	16S	80%, 5T, 5P	0.9615	0.8686–0.9928	0.9069
T1D cohort	WGS	90%, 100T, 10P	0.9481	0.6774–0.9987	0.9036
Antibiotics cohort	16S	70%, 0T, 0P	0.8772	0.8312–0.9285	0.8643
Antibiotics cohort	WGS	80%, 10T, 10P	0.7916	0.6502–0.8951	0.7205

Bold numbers represent highest values for the given data set. 16S RNA and WGS data was tested for each of the three data sets. Optimal model parameters are the values used to obtain the highest accuracy for the data set

ISSN: 1471-2105