Skip to main content

Table 1 Comparisons of 30 scoring functions on the CASF-2013 dataset

From: DLIGAND2: an improved knowledge-based energy function for protein–ligand interactions using the distance-scaled, finite, ideal-gas reference state

Function PCC RMSE Description Year
RF-Score-v2 0.803a 1.54 Machine learning 2014
ID-Score 0.753b 1.63 Descriptor-based and empirical 2013
ΔvinaRF20 0.686c 1.64 Machine learning 2016
AutoDockHybrid 0.64 n.a. Force fields and machine learning 2016
X-ScoreHM 0.614 1.78 Empirical 2002
ΔSASA 0.606 1.79 Empirical 2014
ChemScore@SYBYL 0.592 1.82 Empirical 1998
ChemPLP@GOLD 0.579 1.84 Empirical 2009
DLIGAND2 0.572 1.85 Knowledge-based This paper
SMoG2016 0.57d 1.68 Knowledge-based and empirical 2016
PLP1@DS 0.568 1.86 Empirical 2000
AutoDock Vina 0.563e 1.87 Knowledge-based and empirical 2010
G-Score@SYBYL 0.558 1.87 Energy-based 1997
ASP@GOLD 0.556 1.88 Statistical potential 2005
ASE@MOE 0.544 1.89 Empirical n.a.
ChemScore@GOLD 0.536 1.90 Empirical 2003
DLIGAND 0.526 1.92 Knowledge-based 2005
D-Score@SYBYL 0.526 1.92 Energy-based 2001
Alpha-HB@MOE 0.511 1.94 Empirical n.a.
LUDI3@DS 0.487 1.97 Empirical 1998
GoldScore@GOLD 0.483 1.97 Energy-based 1997
Affinity-dG@MOE 0.482 1.98 Empirical n.a.
LigScore2@DS 0.456 2.02 Empirical 2005
GlideScore-SP 0.452 2.03 Energy-based 2006
SMoG2001 0.418 3.39 Knowledge-based 2001
Jain@DS 0.408 2.05 Empirical 2006
PMF@DS 0.364 2.11 Statistical potential 2006
GlideScore-XP 0.277 2.18 Energy-based 2004
London-dG@MOE 0.242 2.19 Empirical n.a.
PMF@SYBYL 0.221 2.20 Statistical potential 1999
  1. The results for 23 scoring functions were collected from Li [5], the results for RF-score-v2, ID-score, ΔvinaRF20 and SMoG2016 (labeled as a, b, c, d) were collected from Ballester [57], Li [20], Wang [58] and Theau [31], separately, and the results for DLIGAND2, Autodock Vina, and DLIGAND were calculated with default options by ourselves
  2. n.a. not available