Preparing a Pharmacophore for Virtual Screening with External Applications
LigandScout creates pharmacophore models that can be
used in various external applications for virtual screening. The
following tutorials will show you how to prepare your pharmacophores
to be used in Catalyst, MOE, and Phase.
Preparing a Pharmacophore for Virtual Screening with Catalyst
Creating a Catalyst compatible Pharmacophore
LigandScout is designed for the retrieval of automatically
created
advanced 3D pharmacophore models
supporting multiple features per heavy atom to broaden the scope of a
single model. However, not all in-silico screening applications
support multiple features located on one heavy atom.
If you are using the software-package Catalyst, it is
recommended to use the
Create Simplified Pharmacophore
(Catalyst)
routine. The process is started by pressing
SHIFT+F9 or using the
Create Simplified Pharmacophore
(Catalyst)
command in the
menu. In this case, LigandScout achieves compatibility by searching
for the best feature for every heavy atom according to a feature
priority. This priority can be adjusted in the
'chemicalFeatureDefinitions.xml'
file.
You may also reduce pharmacophores for Catalyst by manually
deleting features. In this case, it is recommended to first delete
features interacting with water molecules and then to erase
features that have unfavorable distances and angles. Try to
preserve the best feature for every heavy atom if there are
multiple features on one heavy atom. You may also use literature to
sort out interactions.
Exporting and Converting Pharmacophores for Use in Catalyst
Open the
Save
dialog
by selecting
>
in the menu bar. Choose
Pharmacophore as Catalyst Hypoedit Script
(*.hypoedit)
in the file type drop-down menu, enter a
file name and export the pharmacophore. It is highly recommended to
also export the ligand as MDL Mol file for later hypothesis
verification. Herewith, work in LigandScout is finished.
Next, the
.hypoedit
script has to be
converted into a
.chm
hypothesis file, which can be imported into Catalyst as a hypothesis
ready to be used for virtual screening.
The following steps assume that you have properly installed
Catalyst on your workstation. Since LigandScout uses an extended
dictionary, you have to provide this file, that is shipped with
your LigandScout distribution, for the conversion step. As
LigandScout sets the dictionary path in exported
.hypoedit
scripts to
$CATALYST_DICTIONARY/Dictionary.chm
you either have to overwrite your
$CATALYST_DICTIONARY/Dictionary.chm
with the LigandScout version or you use a text editor to adapt the
second row of the
.hypoedit
file (i.e.
-dict $CATALYST_DICTIONARY/Dictionary.chm \
) with respect to the location of LigandScout's
Dictionary.chm
version.
Now that all environmental needs are accomplished, use the
command
chmod a+x filename.hypoedit
to flag your script as executable. Run the script by typing
./filename.hypoedit
. The script starts a short
conversion process resulting in a .chm
file that you can import in
Catalyst.
Importing a Hypothesis into Catalyst
Run Catalyst and import your
.chm
file by selecting in the menu
>
.
Verifying the Hypothesis in Catalyst
It is highly recommended to verify Catalyst's
fitting capability of this hypothesis with the respective ligand.
Only when Catalyst is able to fit ligand and hypothesis, can you be assured that
Catalyst will find the respective ligand in databases. Thus, please
also import the ligand's MDL Mol file into your stockroom. Drag
& Drop both compounds into the
Hypotheses Workbench
.
To open the
Show Conformational Model
dialog,
select in the menu
>
.
Select
conf1
in the
Conformers Panel
and click on
Register Confs
at the bottom of the window.
Now that this conformer is registered, return to your workbench and
ensure that both ligand and hypothesis are still selected. To use the
Compare/Fit
function, choose in the menu
>
.
A new window will pop up on the
left side of your workbench offering several fitting options. For a
first test, try to get a fitting with the defaults. Catalyst will
instantly depict a fitting in most cases. If you are unable to
retrieve a fitting, please see the
guidelines and
suggestions
below for troubleshooting instructions.
Otherwise, try to fit these compounds with
Max. Omitted
Features
set to '0'. This modification causes Catalyst
to consider the fitting of every single feature of the hypothesis.
Fitting will only be successful, if all features can be fitted and
only if this fitting is successful, Catalyst will find this ligand
by database search. In other words, if you want to ensure that your
ligand will be retrieved by Catalyst during a database search, you
have to ensure that Catalyst is able to fit this compound in the
Hypotheses Workbench
with
Max.
Omitted Features
set to '0'. The following list provides
guidelines and suggestions on how to deal with the fitting
issue.
Fitting Hypotheses in Catalyst: Guidelines and Suggestions
It may seem quite complicated to you to fit LigandScout
hypotheses in Catalyst. However, with the following suggestions you
will be able to get fittings of highly selective pharmacophores the
quick and easy way!
-
Make sure that there are no multiple features on one
heavy atom in the pharmacophore. Catalyst does not support multiple
features on the same position in 3D.
-
Set
compare.minInterBlobDistance=0
in
the
.Catalyst
parameters file to
decrease the overall minimum distance between features. Please see
the Catalyst documentation for detailed instructions on the
.Catalyst
parameters
file
-
Set
FunctionMapping.Hydrophobe.Neighbor.?numBondLessEqualFromOWithDoubleBond=0
in the
.Catalyst
parameters file
if you experience fitting problems of hydrophobic features e.g.
with sulfonamides.
-
Try to get a fitting with
Max. Omitted
Features
set to '1' first, and only if this
works, with the parameter set to '0'
-
Locate the problematic pharmacophore feature(s): With
Max. Omitted Features
set to '1',
analyze the fitting. Fitted features with projection point (e.g.
hydrogen bond features) show a vector. This vector lacks in omitted
features! In omitted spheres of all feature types the central,
small, light-blue sphere is missing! Excluded volumes are not
problematic for the fitting process and can be
disregarded.
-
Hydrophobic features are most likely to be omitted by
Catalyst's fitting algorithm. Special care should be taken on
methyl moieties and chains.
-
Watch out for very tight feature distances, e.g., at
heterocyclic systems. These moieties tend to have a lot of features
focused in small space, which can be an issue.
-
In some cases, hydrogen bonds are only fitted after
generation of a conformational model, because Catalyst is very
restrictive on interaction angles. Generate conformers to check if
the features get fitted. However, it is recommended to adjust
pharmacophores to be fittable with the bioactive conformation of
the PDB complex.
-
Hydrogen bond feature fitting problems are most likely to
occur at heterocyclic and sterically hindered moieties.
-
Dismount the pharmacophore beginning with suspect
features and always save intermediate steps as local object with
>
.
-
Catalyst may deny fitting hydrogen bond features due to
the location of its projected points. Try to delete this sphere and
the respective vector. Even without direction, these features are
valuable for highly selective pharmacophores.
Adding Features in Catalyst
In special cases you may want to add a feature to a
pharmacophore within Catalyst.
Following guide shows you how to add features by using the
Show Function Mapping
feature, which enables
accurate placement of new features on heavy atoms of the PDB
ligand.
-
Import ligand and hypothesis to your workbench and
prepare the pharmacophore in a way that Catalyst is able to achieve
fitting with
Max. Omitted
Features
set to '1'.
-
Add your favored feature (e.g. a hydrogen bond acceptor)
to the desired heavy atom with Catalyst's
feature located in the
menu.
-
Use LigandScout to retrieve the coordinates, e.g., for the
projected point of your hydrogen bond acceptor feature. Click the
Toggle Tooltips
icon of the LigandScout tool bar. LigandScout shows you the exact
position of the currently selected item in a box adjacent to your
selection. These coordinates can be used to set the location
constraints in Catalyst.
-
Select the respective projected point of the hydrogen
bond feature in your
View Hypotheses
workbench and
alter the coordinates in the
Set Constraint
Tolerance
dialog. The Catalyst documentation
gives you detailed information on how to add features to a
hypothesis.
Preparing a Pharmacophore for Virtual Screening with MOE
Creating a MOE compatible Pharmacophore
LigandScout is designed for the retrieval of automatically
created
advanced 3D pharmacophore models
to broaden the scope of a single
model. However, not all in-silico screening applications support
exactly the same features or define their features in exactly the
same way.
If you are using the software-package MOE, it is highly
recommended to use the
Create Simplified Pharmacophore
(MOE)
routine. The process is started by pressing
CTRL+SHIFT+F9 (Shift+Command+F9 on MAC) or selecting
>
. In this case,
LigandScout achieves compatibility by merging features at the same
position and by adding charge features
if and only
if
one of the underlying atoms have an explicit charge.
Also the hydrogen bond lengths have to be adjusted to be
1.8 Ĺ
.
Exporting and Converting Pharmacophores for Use in MOE
Open the
Save as File
dialog in the
menu. Choose
Pharmacophore in CCG MOE Format (*.ph4)
in the
file type drop-down menu, enter a file name and export the
pharmacophore in one of the two supported MOE pharmacophore
schemes. It is highly recommended to also export the ligand as MDL
.mol file for later hypothesis verification.
Importing a Hypothesis into MOE
Run MOE and import your
.ph4
pharmacophore by selecting in the menu
>
. Choose your
.ph4
file, select
moe_ph4
in the drop-down box next to
Force type
and
click
OK
.
Verifying the Hypothesis in MOE
It is recommended to verify the exported pharmacophore
against its originating molecule in MOE. To do this, you first
create a database by selecting in the menu
>
>
. Enter a name for your database
and click
OK
. Now you need to import your
previously exported MDL .mol file into the newly created database.
Select
>
from the
menu. This will open a new dialog. Click
on
Add...
, replace
*.sdf
by
*.mol
and press
Enter
.
Select your previously exported .mol file and click
OK
. Also click
OK
in the
Import SD files
dialog. Your molecule should
now be in the database. However, a final step is required to match
LigandScout's features: Select
>
and choose
PHCD
in the drop-down menu next to
Scheme
. Click on
OK
and
your database is all set.
Now, click on
in the
Pharmacophore Query Editor
. In the
Open
dialog choose your newly created database (*.mdb) and click
OK
. Click on
Search
and
if everything worked out, you will see a status message saying "1
molecule, 1 hit". Otherwise, at least one feature did not match and
needs some work. The check box
Enable Partial
Match
in the
Pharmacophore Query
Editor
allows you to find out how many features did not
match by iteratively decreasing the number of features which need
to match.
Preparing a Pharmacophore for Virtual Screening with Phase
Creating a Phase 2.0 compatible Pharmacophore
LigandScout is designed for the retrieval of automatically
created
advanced 3D
pharmacophore models
to broaden the scope of a single
model. However, not all in-silico screening applications support
exactly the same features or define their features in exactly the
same way.
If you are using the software-package Phase, it is
recommended to use the
Create Simplified Pharmacophore
(Phase)
routine. The process is started by pressing
CTRL+SHIFT+F10 (Shift+Command+F10 on MAC) or using
>
. In this case, LigandScout achieves compatibility by reducing multiple
H-acceptor or donor features at the same ligand atom to a single
feature of the respective type, and by implementing the hydrophobic
feature perception algorithm of Phase. Hydrogen bonding
interactions are exported as point features to overcome the lack of
freely choosable bonding directions in combination with projected
points and Phase's mandatory placement of H-donor features at the
position of hydrogen atoms.
You may also reduce pharmacophores for Phase by manually
deleting features. In this case, it is recommended to first delete
features interacting with water molecules and then to erase
features that have unfavorable distances and angles. Try to
preserve the best feature for every heavy atom if there are
multiple features of the same type on one heavy atom. You may also
use literature to sort out interactions.
Exporting Pharmacophores for Use in Phase
Open the
Save
dialog in the
menu. Choose
Pharmacophore as Phase Files (*.xyz *.xvol *.def *.dxyz
*.mae)
in the file type drop-down menu, enter a file
name and export the pharmacophore. It is highly recommended to also
export the ligand in a format readable by Phase for later
hypothesis verification. Herewith, work in LigandScout is
finished.
Importing a Hypothesis into Phase
Run Maestro and import your hypothesis
.xyz
file using the
Hypotheses
Table
dialog reachable from the
menu. Afterwards, you can
see the pharmacophore and excluded volumes as colored balls in the
Maestro main window.
Verifying the Hypothesis in Phase
The following verification procedure assumes that the ligand
structure has been exported from LigandScout as
ligand.sdf
and
the corresponding Phase hypothesis has been saved under the name
hypothesis.xyz
. Furthermore, the hypothesis and ligand files have
to be accessible from the current working directory.
-
Create a Phase database containing only the
ligand:
$SCHRODINGER/utilities/phasedb_manage -new sdf -db
`pwd`/ligand_db -sd ligand.sdf -confs true cp hypothesis.def
ligand_db_feature.ini $SCHRODINGER/utilities/phasedb_confsites -db
`pwd`/ligand_db -JOB create_ligand_db
-
Try to match the hypothesis with the ligand
molecule:
$SCHRODINGER/utilities/phasedb_findmatches -setup hypo_search
-db `pwd`/ligand_db -hypo `pwd`/hypothesis -mode find+fetch
-deltaDist 2 -useExistingSites true -useDbKeys false
$SCHRODINGER/phase_dbsearch hypo_search
Successful execution of the above commands results in a
search log-file called
hypo_search_dbsearch.log
.
Status messages near the bottom of the file should indicate a successful overall
search procedure and that 1 match to the hypothesis has been found
in the database.
|