Table 6.

ARG-SHINE outperforms existing ARG classification programs and the component methods in terms of classification mean accuracy and weighted-average F1-score on the COALA100 test data

Methods	Accuracy	Macro-average F1-score	Weighted-average F1-score
BLAST best hit	0.9066	0.9131	0.9221
TRAC	0.9043	0.8963	0.9020
ARG-CNN	0.9219	0.9176	0.9206
ARG-InterPro	0.8689	0.8135	0.8654
ARG-KNN	0.9112	0.9215	0.9266
ARG-SHINE	0.9286	0.9225	0.9276

The best results among all the methods and best results among the stand-alone methods are in bold.