® : Electronic High Throughput
Enrichment Experiments for eHiTS
eHiTS 2009 Enrichmet technical note.
To see other eHiTS 2009 technical notes, visit our recently updated technical notes pages.
eHiTS preforms very well in a screening evaluation by Merck
eHiTS was recently evaluated on 11 targets of pharmaceutical interest in collaboration with Merck Research Laboratories. The original study (without eHiTS) was published by Dr. Georgia McGaughey et.al. in 2007 .
This recent testing of eHiTS on the same set revealed that eHiTS overall performed very well across those 11 families. In particular, it outperformed all the other programs tested on the Carbonic Anhydrase I family (pdb code: 1azm) with a RIE(*) value of 51.8. The eHiTS performance average for the entire test set was 18.7, which was the highest average amongst all the tested programs, that included Glide, Fred and Flog. Also the sum of RIE's over all 11 targets is the highest for eHiTS, as shown in the chart below.
Note (*). RIE stands for Robust Initial Enhancement and was proposed as a metric to
evaluate the performance of ranking methods in virtual screening by Sheridan et al , and recently formulated with an analytical formula by Truchon & Bayly .
[ref 1] Comparison of Topological, Shape, and Docking Methods in Virtual Screening
G.B. McGaughey, R.P. Sheridan, C.I. Bayly, C. Culberson, C. Kreatsoulas, S. Lindsley, V. Maiorov, J. Truchon and W.D. Cornell
J. Chem. Inf. Model. 2007; 47(4), pp 1504 - 19 DOI: 10.1021/ci700052x
[ref 2] Protocols
for bridging the peptide to nonpeptide gap in topological similarity
Sheridan, R.P.; Singh, S.B.; Fluder, E.M.; Kearsley, S.K.
J. Chem. Inf. Comput. Sci. 2001, 41, 1395-1406.
[ref 3] Evaluating
Virtual Screening Methods: Good and Bad Metrics for the "Early
Jean-François Truchon and Christopher I. Bayly
J. Chem. Inf. Model.; 2007;
47(2) pp 488 - 508; DOI: 10.1021/ci600426e
Results for eHiTS v5.1
eHiTS 5.1 ships with scoring functions that have been
trained on a wide
range of protein families. eHiTS will perceive which family a
given receptor belongs to and will use the appropriate scoring function
(using the default scoring function when a matching family is not
found). To test the effect of this scoring function on enrichment
values, 5 receptor targets were chosen to represent a variety of
receptor types. Each receptor was identified as belonging to a
trained family set.
- Cox-2 (1CX2 - 14 actives)
- Estrogen (1ERR - 10 actives)
- Factor Xa (1IQN - 15 actives)
- p38-MAP Kinase (1BL7 - 13 actives)
- Neuraminidase (1IVF - 9 actives)
Cox-2 and Estrogen have lipophilic buried cavities, while Factor Xa and
p38-MAP Kinase have active sites with intermediate polarity with
hydrogen bonding motifs common to the majority of inhibitors.
Neuraminidase has a very polar, solvent-exposed binding sites.
For these runs, 1000
randomly selected MDDR drug-like ligands were used
as the decoys. These decoys were not screened to remove any
The above plot shows that eHiTS with the default trained scoring
function gives very strong enrichment results for all the receptors