Biology Reference
In-Depth Information
Table 2.3
Number of hit proteins in each pocket prediction class on the drug-target dataset
for each method
Method
1st pocket
2nd pocket
3rd pocket
None
MetaPocket
121
17
9
51
LIGSITE
CS
95
18
7
78
PASS
69
30
11
88
Q-SiteFinder
79
28
16
75
SURFNET
46
11
8
133
GHECOM
78
22
10
88
ConCavity
93
12
6
87
Fpocket
61
34
17
86
POCASA
83
23
4
88
on surface generally and obviously there is no strong correlation between the number
of cavities and the prediction success rate of metaPocket.
It is believed that ligands trend to binds to the large pocket site on protein
surface. In order to check whether ligands bind to large pockets on protein surface,
we conducted a statistical analysis to assess the possibility that a real ligand-binding
site locates at the top 3 identified pockets. Here the identified pocket sites are
classified into four different classes: the actual ligand binding site locates at the first,
the second, the third pocket, or at none of these top 3 pockets (Table
2.3
). In the top
3 predictions of metaPocket, there were 121 (61%) cases that the top-1 predicted
pocket is the real ligand-binding site. There were 17 and 9 cases that the second,
the third prediction pocket is the real ligand-binding site, respectively. However,
there were 51 cases for which the metaPocket failed to detect the real ligand-binding
site (RBS) among the top 3 predictions. Among the 121 cases that ligands were
predicted to bind to the first pocket site in metaPocket, in 94 (78%) cases, the
predictions overlap with one of the top 3 identified pockets identified by all of the 8
single methods and in 17 (14%) cases the predictions overlap with one of the top 3
identified pockets identified by 7 out of the 8 single methods. Only in 12 of the 121
cases, the real-ligand binding sites were predicted by all 8 single methods at the
top-1 prediction.
2.6
Conclusion
To make LIGSITE
csc
and metaPocket available to the community, we built easy-
to-use web-servers and make them online at
http://projects.biotec.tu-dresden.de/
pocket/
and
http://projects.biotec.tu-dresden.de/metapocket
. Generally it only takes
several seconds for LIGSITE
csc
to finish pocket identification, depending on the size
of protein. In metaPocket, eight single methods are called in parallel to reduce
computational time. Each of eight single methods is treated as a plug-in and thus it
is very easy to add other new methods into metaPocket, to further improve ligand
Search WWH ::
Custom Search