ScreeningAssistant
Manual
More scientific details about the software can be found
here.
NEEDED
SOFTWARES
Before you can use ScreeningAssistant, 3 softwares must be installed on
your computer:
- Java
5 JRE (ScreeningAssistant won't
work with previous Java virtual
machine)
- Java3D
(Java3D must be installed AFTER Java 5 JRE)
- A MySQL server to manage
databases. It is
recommended to run MySQL on a separated Linux computer. However, for
testing
purpose, you can use EasyPHP.
EasyPHP is a Windows software which automatically installs a MySQL
server on
your computer.
- Optionally, you can use Marvin
as molecular viewer by adding MarvinBeans.jar to your classpath. If you
don't have Marvin, ScreeningAssistant will use its internal molecular
viewer (JOELib molecular viewer).
IN CASE OF
PROBLEMS
The current version has been tested on windows
2000/XP/vista . If it doesn't work on your computer please carrefully check
your installation
You cannot obtain the first screen ( the
one on the image 10 lines lower):
- Did you correctly install
Java 5 JRE ( caution: JRE, not JDK)
- Did you install Java3D
after Java JRE
You cannot connect to MySQL server:
- Is the server running (
either on your computer( 127.0.0.1) with easyphp or on another server?
- If the server is on
another computer, most of the problems are coming from the firewalls
Everythings seems OK but you cannot import
SDF files:
- Did you try a small SFD
file ? With big ones, the importation step may be very long ( nothing appears
before the end of the import step)
FOR USERS OF A PREVIOUS VERSION OF SCREENINGASSISTANT
This new version introduces changes in the structure of the database
used by ScreeningAssistant. In consequence you can't use this new
version directly with a database created with a previous version of
ScreeningAssistant. Your old databases need to be updated. It could be
done using a little java program. Here is an example for updating a
database named MyDataBase:
In this example the MySQL server is installed on this computer, the user name is Toto and his password is MyPassword.
USER
GUIDE
The
ScreeningAssistant desktop environment: create your first database
Launch your MySQL server (e.g. easyPHP) and double-click on the
ScreeningAssistant icon on your desktop.
(You must launch EasyPHP each time you want to use ScreeningAssistant.
EasyPHP launch two softwares: Apache (a web server) and MySQL. Apache
don't need to be started, only MySQL wil be used by ScreeningAssistant.)
The following windows will
appear on the screen :
You can see that only an internal window is open, the molecule viewer,
which is
empty. ScreeningAssistant GUI is build as a virtual desktop, and it
means that you can minimize, resize and move the opened windows. First
thing to do is to open an existing database or to create a new one. To
do this open the Database
menu and click on New.
In the login window, enter the server's IP address (127.0.0.1 if you
use easyPHP on your computer), a user name of a MySQL administrator
account and the corresponding password (by default MySQL admin name is
"root" with no password).
Clicking OK
in the
login window will let you create a new database or open an
existing one. To create a new database, enter the name in the text box
and click create. To open an existing database select it and click Open.
You can also delete a
selected database.
When a database is opened, you can see its compounds in the Database
Viewer window. Gey rows means that corresponding molecules are
estimated as not drug-like. Clicking in any row of the database viewer
will display the molecule in the structure viewer.
At your first use you won't have any row in your Database Viewer. The
next step will be to import a SDF.
The
menu bar: import/export compounds and analyse your database
To append compounds in the database go to the Database
menu of the Database Viewer window, and click Append.
In
the Append window, you must first choose the file to import. if "Add
all SDF of this directory" is set to Yes, you must be sure that all the
SDF of the directory have the same ID and CAS field names.
Then, you must choose which field names corresponds to ID field and CAS
field. CAS field is optional, but you must define ID field. In the
provider part, you can choose the provider of the compounds, either an
existing one or a new one (you must then define at least the Name
field). Click OK
to begin the importation process.
To export the selection of the Database Viewer (we will explain how to
change the selection in the tool bar part of this manual), click on
Export
in the
Database
menu. In the
export window, you can define name
of the exported file and its type. If you have minimized structures
with Corina, you can choose to export only correctly minimized
structures.
If a number is entered in
the field "Diversity: number of compounds to select", then a diversity
algorithm will be used to select a number of compounds close to the
number entered (the number of compounds exported by diversity can't be
specified exactly). Diversity algorithm is quick for a small number of
compounds, but can be very long for million structures !
Many charts are available in ScreeningAssistant :
- Pie
Charts :
compounds by
providers
- Compounds Repartition
- Drug-like
- Lead-like
- Percentages
- Distribution
- Molecular Weight
- Number of H acceptors
- Number of H donors
- LogP
- Number of rotatable bonds
- TPSA
- Internal
Duplicates
- Unicity:
compound not present in the databases of the other providers
- Frameworks:
number of different frameworks by
providers
- Diversity:
estimated with the number of clusters created by the SCA algorithm with
0.8 cut-off
- Global
- Drug-like
- Lead-like
- Chemical
Space:
display selected compounds in a chemical space. The axis are defined by
the PCA1 and PCA2 axis equations obtained from the PCA analysis of 18
000 compounds selected by diversity among 2.6 millions.
At
the present time, only the minimization of the compounds by Corina
is available. Of course you need the corina.exe file in your computer
to use this option. Once corina.exe put in the
ScreeningAssistant\external\corina directory, you can minimize
compounds of your databases.
The
tool bar: navigate in your database and select compounds
The database viewer displays 1000
molecules by page. You can change the pages using the five blue icons.
The 
icon allows to directly go to a
page
entering is
number.
You can select desired compounds using the 
icon.
You can select the providers. By default all providers are selected. In
the query part of the screen you can define filters for the selection.
You must choose the parameter, the equality sign, the value and then
click Add
to define a
new parameter. There is no limit in the number of filters you can
define.
Parameters:
- id:
internal ID
of the compound
- md5ichi:
hascode
of the unique identifier
- mw:
molecular
weight of the compounds, without counter-ion(s) and at pH=7
- logP
- TPSA:
Topological Polar Surface Area
- Hba:
number of H
bond acceptors
- Hbd:
number of H
bond donors
- rotatable
bonds
- halogens
- single_chains:
single chains longer than ≤
-(CH2)6CH3
(0 for no, 1 for yes)
- perfluorinated_chain:
presence of -CF2CF2CF3
(0 for no, 1 for yes)
- SSSRs:
number of
Smallest Sets of Smallest Rings
- big_ring_size:
size of the bigger ring found by SSSRs
- O:
number of O
atoms
- N:
number of N
atoms
- S:
number of S
atoms
- NO2:
number of NO2
- SO2:
number of
SO2
- CF3:
number of CF3
- CF3_halogens:
number of CF3
and other halogen atoms
- NOS:
number of
N, O and S atoms
- bad_atoms:
number of atoms other than C, O, N, S, P, F, Cl, Br, I, Na, K,
Mg, Ca, or Li
- is_reactive:
presence of a reactive function (0 for no, 1 for yes)
- is_warhead:
presence of a warhead type substructure (0 for no, 1 for yes)
- is_promiscuous:
the molecule is known to be a promiscuous inhibitor (0 for no, 1 for
yes)
- is_privileged:
if > -1, then a privileged structure is detected in the
structure of
the molecule
- absorption:
not
used yet
- caco2:
not used
yet
- solubility:
not
used yet
- BBB:
not used yet
- drug-like_failures:
number of non-fitted drug-like criteria
- lead-like_failures:
number of non-fitted lead-like criteria
- PDL_score:
our internal Progressive 'Drug-Like' score. ≤
1 means the compounds is estimated drug-like, more than 2 means the
compounds is absolutely not drug-like
- PLL_score: our
internal Progressive 'Lead-Like' score. ≤
1 means the compounds is estimated lead-like, more than 2 means the
compounds is absolutely not lead-like
- CFMS:
Cleaning For My Screening. By default it is based on PDL and add
penalties for the presence of reactive functions, warheads, primiscuous
aggregating inhibitors, single chains, perfluorinated chains, and for
the abscence of N or O
- scaffold:
scaffold id (scaffold used is rings + linkers)
- framework:
id of
the general 2D shape of the molecule
- ClogP:
not used yet.
- entrie_date
Example
of a drug-like selection:
Example
of a selection for screening:
A screening selection can be done in the same way that a drug-like
selection, but using the criteria bad_atoms=0 and CFMS=0. By default,
CFMS is the same thing that PDL, but it takes into account additional
features such as reactive functions.
CFMS can be personalized using the
icon. Then, you can choose
the base of CFMS ('drug-like' or 'lead-like'
i.e. PDL or PLL)
the additional penalties can be chose. Clicking
Ok after
user's choices will recompute the new CFMS value for all the compounds
of the current database.
The reactive functions and warheads penalties can also be personalized
through
Configure
in the main menu of ScreeningAssistant :
The False Positive Filters window is now open:
This window will allow you
to modify definitions of reactive functions and warheads. You
can create and delete sub-structures filters using
New and
Delete
buttons. The name of a property can be changed double clicking on it.
The
Structure
field let you to edit the
SMARTS
code of the filter. If you have MarvinBeans.jar in your
classpath, the sub-structure can be edited graphically.
The modification of the filters will apply when new compounds are added.
icon. Here is an example of chemical space:
This chemical space use the PCA1 and PCA2 axis equation computed on 18
000 diverse molecule selected from 2.6 millions with the descriptors
used are SSKey3DS, MW and logP.
