Skip to main content

Advertisement

Table 5 Selected variables important in statistical models of Set A and Set B.

From: An investigation into pharmaceutically relevant mutagenicity data and the influence on Ames predictive potential

Variable Set A PC1 Mean Decrease in Gini log p Set B PC1 Set B PC2 Mean Decrease in Gini log p
chi0v_C 0.153 (1) 0.4 (30) -5 0.148 (2) -0.070 (86) 1.5 (17) -8
GCUT_SLOGP_3 0.151 (4) 1.07 (1) -4 0.141 (5) -0.059 (102) 2.3 (8) -8
a_count 0.152 (2) 0.3 (62) -6 0.129 (19) -0.108 (51) 1 (54) -2
balabanJ -0.097 (60) 0.5 (15) -2 -0.128 (20) 0.004 (241) 2.6 (7) -8
a_nH 0.149 (5) 0.4 (40) -4 0.115 (35) -0.077 (76) 1 (52) -1
Density -0.108 (51) 0.4 (32) -2 -0.106 (47) -0.058 (103) 2.7 (6) -7
NitFormE -0.125 (26) 1.0 (2) -3 -0.104 (52) -0.04 (136) 8.9 (1) -8
Q_VSA_POS 0.129 (24) 0.6 (7) -3 0.061 (87) -0.052 (108) 1.4 (26) -1
BCUT_SMR_3 0.124 (28) 0.6 (9) -2 0.046 (111) -0.129 (32) 1.4 (24) -1
a_nC 0.146 (10) 0.3 (57) -4 0.148 (3) -0.073 (80) 0.8 (74) -12
Morgan0672 0.056 (111) 0.1 (130) 0 0.107 (44) -0.045 (123) 0.3 (144) -3
a_hyd 0.140 (15) 0.2 (121) -3 0.143 (4) -0.065 (94) 0.5 (114) -5
b_1rotN 0.075 (82) 0.05 (187) -1 0.001 (244) -0.154 (11) 0.3 (146) -4
KierFlex 0.085 (73) 0.3 (69) 0 -0.011 (206) -0.184 (4) 1.6 (14) -7
b_rotN 0.065 (97) 0.03 (217) -1 -0.004 (232) -0.16 (8) 0.3 (146) -5
Morgan0280 -0.042 (147) 0.02 (242) -1 -0.06 (88) -0.149 (14) 0.2 (160) -8
a_nO -0.04 (150) 0.03 (223) 0 -0.073 (75) -0.139 (25) 0.3 (148) -8
  1. PLS loadings, the PLS jackknife significance estimate from leave-one-out cross-validation, and the random forest mean decrease in Gini coefficient for each variable is given below with their overall rank by that measure. All variable abbreviations follow those in MOE. Bolded, italicized, and underlined text respectively indicates variables chosen to be important in both Set A and Set B from the first principal component, important for Set B in the first component, and important for Set B in the second component.