HIV-1 coreceptor usage prediction without multiple alignments: an application of string kernels

Retrovirology

Table 5 Available methods. The results column contains the metric and what the classifier is predicting.

Reference	Learning method	Training set	Testing set	Multiple alignments	Results
Pillai et al. 2003	Charge rule [5, 6]	271	-	yes	Accuracy (CXCR4): 87.45%
Resch et al. 2001	Neural networks	181	-	yes	Specificity (X4): 90.00%
Pillai et al. 2003	SVM	271	-	yes	Accuracy (CXCR4): 90.86%
Jensen et al. 2003	PSSM¹	213	175	yes	Specificity (CXCR4): 96.00%
Jensen et al. 2006	PSSM	279	-	yes	Specificity (CXCR4): 94.00%²
Sander et al. 2007	SVM	432	-	yes	Accuracy (CXCR4): 91.56%
Xu et al. 2007	Random forests	651	-	yes	Accuracy (R5): 95.10%
Lamers et al. 2008	Neural networks	149	-	yes	Accuracy (R5X4): 75.50%
This manuscript	SVM	1425	1425	no	Accuracy (CXCR4): 94.80%

ISSN: 1742-4690