[Product Releases]
|
|
|
|
|
[Blog]
|
|
Most recent post
|
[News]
|
|
Can we trust docking results?
Sept 2010 IBM Systems and Technology Group releases a white paper with eHiTS and Cell
Oct 2008
EPA's ToxCastTM project will use SimBioSys' eHiTS as docking engine
Nov, 2007
|
[Events]
|
| 243rd ACS
Mar 25-29, 2012
San Diego, CA
see >> more
|
|
|
|
|
eHiTS LASSO:
Ligand Activity in Surface Similarity Order
|
 |
eHiTS LASSO: results,
trained on just 2% of the actives
|
Following the ligand diversity classification of Hert
et. al J. Chem. Inf Model, 46, 462-470, 2006, three groups of ligands
with low, medium and high Mean Pairwise Similarity (MPS) were extracted
from the MDDR database. For each of these 3 groups 5 families, i.e. 15
sets in total were chosen. Please see detailed results in the table
below or a quick view on the chart above.
Data specifications:
|
Activity class
|
Activity Key
|
# Compounds
|
Diversity (MPS)*
|
% recovered in top 2%
|
% recovered in top 5%
|
% recovered in top 10%
|
|
High Diversity
|
|
nitric oxide synthase inhibitors
|
12464
|
377
|
0.19
|
25.37%
|
37.07%
|
48.78%
|
|
muscarinic (M1) agonists
|
9249
|
848
|
0.21
|
45.02%
|
63.41%
|
70.69%
|
|
aromatase inhibitors
|
75721
|
513
|
0.23
|
24.76%
|
36.43%
|
52.38%
|
|
dopamine -hydroxylase inhibitors
|
31281
|
95
|
0.23
|
65.33%
|
92.00%
|
98.67%
|
|
aldose reductase inhibitors
|
43210
|
882
|
0.23
|
21.02%
|
29.30%
|
38.24%
|
|
Medium Diversity
|
|
5 HT2B antagonists
|
6249
|
90
|
0.31
|
31.67%
|
78.33%
|
91.67%
|
|
carbonic anhydrase inhibitors
|
16200
|
255
|
0.31
|
39.18%
|
47.94%
|
61.34%
|
|
protease inhibitors
|
78330
|
574
|
0.31
|
10.85%
|
13.45%
|
18.87%
|
|
CRF antagonists
|
6215
|
254
|
0.32
|
46.52%
|
60.43%
|
77.83%
|
|
thrombin inhibitors
|
37110
|
803
|
0.32
|
19.62%
|
34.72%
|
35.85%
|
|
Low Diversity
|
|
cephalosporins
|
64200
|
1312
|
0.5
|
72.90%
|
85.33%
|
90.25%
|
|
adenosine (A1) agonists
|
7707
|
88
|
0.52
|
94.20%
|
100.00%
|
100.00%
|
|
adenosine (A2) agonists
|
7708
|
71
|
0.54
|
98.25%
|
98.25%
|
98.25%
|
|
monocyclic -lactams
|
64100
|
76
|
0.55
|
97.10%
|
98.55%
|
100.00%
|
|
vitamin D analogues
|
75755
|
279
|
0.57
|
80.41%
|
87.11%
|
94.33%
|
|
Cleaned MDDR
|
|
Cleaned MDDR
|
78050
|
|
0.200
|
|
|
|
* Diversity measurements come from Hert's paper and were
calculated on a slightly different dataset. Our dataset used the same
activity classes, however duplicate structures were removed used our
own feature vector measurements, which will remove structures that
may not be identical but which produce the same feature vector. This
is the reason for the reduced data size.
Data preparation:
For each molecule in the MDDR database, the feature
vector was
calculated. Any duplicate feature vectors were removed. This was to
remove any duplicate molecules and to ensure a dataset that was
usable in the LASSO testing experiment. Each set of ligands were
extracted from the cleaned MDDR database and used in the experiments.
2% of the actives for each set were randomly selected and used to
train the eHiTS LASSO neural nets. The trained neural nets were
used to screen the full cleaned MDDR database.
|
|
|
