Table 2 Prediction performance of co-receptor usage models.

Subtype	Method	Sens	Spec	Acc	PPV	NPV	FPR	FDR	F1	DOR
C	T-CUP	91.07	98.60	97.59	91.07	98.61	1.39	8.93	91.07	513
	g2p	87.50	97.77	96.39	85.96	98.04	2.23	14.04	86.73	244
	PhenoSeq	91.38	92.48	92.33	75.71	98.56	4.74	24.29	82.81	172
	X4R5	75.00	94.49	91.89	67.74	96.08	5.51	32.26	71.19	47
	SINSI	71.43	98.90	95.23	90.91	95.73	1.10	9.09	80.00	174
	SINSI.C	89.29	91.46	91.17	61.73	98.22	8.54	38.27	72.99	76
	Raymond	89.29	99.45	98.09	96.15	98.37	0.55	3.85	92.59	879
A	T-CUP	18.18	96.32	55.39	84.44	51.69	3.68	15.56	29.92	5
	g2p	15.79	97.89	54.89	89.19	51.38	2.11	10.81	26.83	7
	Phenoseq	17.70	94.74	54.39	78.72	51.14	5.26	21.28	28.91	4
	X4R5	15.31	93.94	53.56	72.73	51.24	6.06	27.27	25.30	3
	SINSI	11.54	97.98	53.69	85.71	51.32	2.02	14.29	20.34	5
	SINSI.C	37.80	58.59	47.91	49.07	47.15	41.41	50.93	42.70	1
	Raymond	11.00	98.48	53.56	88.46	51.18	1.52	11.54	19.57	6
	Esbjörnsson	13.40	99.49	55.28	96.55	52.12	0.51	3.45	23.53	16

For subtype C each algorithm achieved a sensitivity of around 90%. Prediction performance for subtype A generally resulted in sensitivities lower than 20%.

Quick links

Search