Skip to main content
. Author manuscript; available in PMC: 2018 May 9.
Published in final edited form as: J Chem Inf Model. 2017 Aug 7;57(8):1859–1867. doi: 10.1021/acs.jcim.6b00694

Table 3b.

Target classes for which SVM shows the most significant changes in classification performance (ΔpAUC) and error (ΔεAUC) relative to LASSO classification for standard ECFP4 fingerprints.

Target (ID) ΔpAUC ΔεAUC z-score
GCPR44 (20174) 0.39% −97.1% 1.32
c-SRC (10434) 0.86% −92.0% 1.31
HERG (165) 0.58% −41.5% 1.26
MAPK14 (10188) 0.40% −72.1% 1.24
D4 (90) 0.63% −67.1% 1.21
D2 (72) 0.30% −68.7% 1.19
NET (100) 0.46% −58.5% 1.19
COX2 (126) 0.42% −71.8% 1.12
NPY5R (11336) 0.15% −98.1% 1.10
ACHE (93) 0.47% −69.1% 1.09