Table 1. Performance evaluation and comparison of the GPS-ARM with known motifs.

Method	Threshold	Ac(%)	Pr(%)	Sn(%)	Sp(%)	MCC
D-box (GPS-ARM)	High	87.29	82.46	63.51	95.39	0.6463
	Medium	80.76	61.54	64.86	86.18	0.5018
	Low	76.63	53.26	66.22	80.18	0.4346
	e	81.10	95.24	27.03	99.54	0.4471
KEN-box (GPS-ARM)	High	86.67	100.00	81.82	100.00	0.7385
	Medium	91.67	100.00	88.64	100.00	0.8218
	Low	95.00	100.00	93.18	100.00	0.8858
Motif-D1a		77.32	90.00	12.16	99.54	0.2797
Motif-D2b		75.95	56.25	24.32	93.55	0.2488
Motif-D3c		76.29	53.25	55.41	83.41	0.3832
Motif-KENd		38.33	100.00	15.91	100.00	0.2192

For the construction of the GPS-ARM software package, the three thresholds of high, medium and low were selected for D-box and KEN-box, respectively.

^aMotif-D1, RXXLXX-I/V-XN [3];

^bMotif-D2, RXXLXXXXN [4], [5], [16];

^cMotif-D3, RXXLXX-L/I/V/M [30];

^dMotif-KEN, KENXXX-N/D [3], [22];

^eFor comparison, we fixed the Sp value of GPS-ARM so as to be identical with Motif-D1.