 |
|
| |
Glide - Case Study 2
Glide outperforms GOLD and ICM in docking accuracy tests and database screens
In a recent study1 conducted at Vertex Pharmaceuticals, Glide, GOLD, and ICM were compared in head to head docking accuracy tests and database screens. Glide's superior performance across a broad range of targets led researchers to conclude that Glide combines the highest degree of docking accuracy with an empirical scoring function that consistently performs well. |
|
 |
To assess docking accuracy, the researchers selected 150 protein-ligand complexes, representing 63 pharmaceutically relevant targets. The structures, taken from the PDB and an in-house collection, were exhaustively culled to maximize diversity while eliminating complexes without drug-like ligands. Glide, GOLD, and ICM were then used to dock each ligand into its corresponding active site.
The authors found that Glide identified a top-ranked structure within 2.0 angstroms of the crystal structure 61% of the time, while top-ranked structures returned by GOLD and ICM met this criterion 48% and 45% of the time, respectively.
After performing a statistical analysis of docking accuracy with regard to structural descriptors, the researchers concluded that Glide was the least sensitive to variations in ligand flexibility, prevalence of hydrophobic or hydrogen-bond interactions, and the  binding site's degree of solvent exposure. Glide's ability to produce accurate poses when GOLD and ICM could not led the researchers to conclude that Glide offers more extensive coverage of conformational space.
Scoring functions were evaluated in virtual screens on HIV-1 protease, IMPDH, and p38 MAP kinase. Glide, GOLD, and ICM were used to generate docked poses for each screen, and the poses were then scored using GlideScore, ChemScore, and the OPLS-AA interaction energy. GlideScore yielded the best overall performance for IMPDH and p38 MAP kinase, and Glide was able to produce docked poses that led to superior results for all screens. Glide's enrichment factors led the authors to conclude that the program is equipped with a capable and effective scoring function that offers consistent performance across several types of binding sites.
The overall results in a direct comparison between Glide, GOLD, and ICM invite the conclusion that Glide features the best performance across the broadest variety of systems, with exceptional accuracy that "makes it the tool of choice in most cases."
|
Enrichment factors calculated on top 3% of ranking for
all pose generation (docking) and scoring combinations
|
|
Docking Method
|
|
P38 MAP Kinase
|
HIV-1 Protease
|
|
Chem-
Score
|
Glide-
Score
|
OPLS-AA
|
Chem-
Score
|
Glide-
Score
|
OPLS-AA
|
Chem-
Score
|
Glide-
Score
|
OPLS-AA
|
|
Glide
|
6.3
|
16.4
|
5.4
|
5.7
|
8.8
|
5.8
|
10.5
|
3.7
|
1.5
|
|
GOLD
|
3.8
|
2.3
|
0.9
|
4.6
|
8.5
|
5.7
|
8.3
|
1.1
|
0.8
|
|
ICM
|
4.7
|
14.1
|
1.4
|
2.2
|
9.0
|
9
|
9.9
|
2.4
|
0.3
|
1 A Detailed Comparison of Current Docking and Scoring Methods on Systems of Pharmaceutical Relevance, PROTEINS, Perola, E.; Walters, W.P.; Charifson, P.S., 2004; 56; 235-249. |
|
