Skip to main content

Table 5 Positive likelihood ratios for HIV-1 PR rules.

From: How to find simple and accurate rules for viral protease cleavage specificities

Rules used Positive likelihood ratio Sensitivity Specificity
HIVA1 11.3 31% 97%
HIVA1, HIVB1 6.2 41% 93%
HIVA1, HIVB1, HIVC1 5.2 57% 89%
HIVA1, HIVB1, HIVC1, HIVD1 4.7 61% 87%
HIVA1, HIVB1, HIVC1, HIVD1, HIVE1 4.7 64% 86%
HIVA2 11.3 33% 97%
HIVA2, HIVB2 6.8 40% 94%
HIVA2, HIVB2, HIVC2 5.6 44% 92%
HIVA2, HIVB2, HIVC2, HIVD2 5.2 48% 91%
HIVA2, HIVB2, HIVC2, HIVD2, HIVE2 4.5 49% 89%
Kontijevskis et al. [17] Fig. 1a 2.1 3% 99%
Kontijevskis et al. [17] Fig. 1b 1.7 87% 49%
Kontijevskis et al. [17] Table III 0.8 3% 97%
Kontijevskis et al. [17] Table IV 1.7 87% 49%
HIVcleave [37] 45.4 2% 100%
HIVcleave [37] 3.0 32% 89%
HIVcleave [37] 2.8 50% 82%
L-SVM [38] 29.1 2% 100%
L-SVM [38] 10.6 31% 97%
L-SVM [38] 6.7 50% 93%
  1. The positive likelihood ratios for different rules and predictors on the HIV-1 3261 data set. The results are shown for the OSRE HIV-1 rules and the Kontijevskis et al. rough set theory rules [17], together with some reference values for the HIVcleave web server [37] and a linear support vector machine [38].