Table 1.

Classification models stats. Basic statistics for each classification model. Models were derived with SARS-CoV-2 sequences downloaded on 2021/11/27. Accuracy was calculated as sequences correctly labeled / number of sequences. Multi-class AUC is the mean AUC from all pairwise class comparisons.

Model	Number of classes	Number of trees	Training dataset size	Testing dataset size	Accuracy	Multi-class AUC
GISAID	10	1000	66,126	13,230	0.9777	0.9931
Nextstrain	22	1000	63,972	12,810	0.9952	0.9879
Pango	1437	500	65,467	12,346	0.9656	0.9926