9. Install JClusterService on a UNIX/Linux system (it's not working under windows and not tested under OSX)
9.1. Create a user under which the jclusterservice should execute the commands
and create a JCLUSTERSERVICEROOTDIRECTORY e.g. /home/maspectras/maspectrasjclusterservice where
the user has access right.
If not already done copy and unpack the maspectras.tar.gz again under the home directory
of the new user. This creates a maspectras directory with all the necessary files.
From there copy all files within maspectras/maspectrasjclusterservice to this directory and make sure that
the user can access the folder. e.g.
groupadd maspectras
useradd -d /home/maspectras -g maspectras maspectras
su - maspectras
mkdir -p /home/maspectras/maspectrasjclusterservice
tar -xzf maspectras.tar.gz
cp -r %extracted path%/maspectrasjclusterservice/* /home/maspectras/maspectrasjclusterservice
chown -R maspectras:maspectras /home/maspectras
# (just to be sure that every file is accessible for the service user execute this command as root)
If you have not already switched to maspectras user, become maspectras user (su - maspectras)
9.2. Edit jboss-4.2.2.GA/server/jclusterservice/deploy/JClusterWS.ear/JClusterWSWeb.war/WEB-INF/server-config.wsdd and change to:
<parameter name="attachments.Directory"
value="%installationpath%/jboss-4.2.2.GA/server/JClusterWS/./deploy/JClusterWS.ear/JClusterWS/./deploy/JClusterWS.ear/JClusterWSWeb.war/WEB-INF/attachments"/>
e.g.
/home/maspectras/maspectras/jboss-4.2.2.GA/server/jclusterservice/deploy/JClusterWS.ear/JClusterWSWeb.war/WEB-INF/attachments
If you change the server ports for jclusterservice in bindings.xml, change also the JNDI URL in this file:
<parameter name="jndiURL" value="jnp://localhost:%your_port%"/>
default is:
<parameter name="jndiURL" value="jnp://localhost:21099"/>
9.3. Adapt the path settings to your local installation in jboss-4.2.2.GA/server/jclusterservice/deploy/JClusterWS.ear/JClusterService.jar/usermanagement-utils.properties
sum_users_filepath=JCLUSTERSERVICEROOTDIRECTORY/config/users.xml
e.g.
sum_users_filepath=/home/maspectras/maspectrasjclusterservice/config/users.xml
9.4. Adapt the path settings to your local installation in jboss-4.2.2.GA/server/jclusterservice/deploy/JClusterWS.ear/JClusterService.jar/cluster.properties
9.4.1. The global path settings:
clusterservicedefinition_path=JCLUSTERSERVICEROOTDIRECTORY/config/clusterservice-definition.xml
job_path=JCLUSTERSERVICEROOTDIRECTORY/jobs
jobdefinitions_path=JCLUSTERSERVICEROOTDIRECTORY/job_definitions
e.g.
clusterservicedefinition_path=/home/maspectras/maspectrasjclusterservice/config/clusterservice-definition.xml
job_path=/home/maspectras/maspectrasjclusterservice/jobs
jobdefinitions_path=/home/maspectras/maspectrasjclusterservice/job_definitions
9.4.2. The queueingsystem specific once:
9.4.2.1. Local execution queue
local_execution_job_template_path=JCLUSTERSERVICEROOTDIRECTORY/template/jms-job-definition
e.g.
local_execution_job_template_path=/home/maspectras/maspectrasjclusterservice/template/jms-job-definition
9.4.2.2. Optional if present ... Sun Grid Engine (SGE)
Adapt the paths to the SGE commands
9.4.2.3. Optional if present ... Portable Batch System (PBS/Torque)
Adapt the paths to the PBS/Torque commands
9.5. Install the following applications:
9.5.1. Database repository management initialization:
9.5.2. NCBI-Blast and NCBI-FormatDB
- download from ftp://ftp.ncbi.nih.gov/toolbox/ncbi_tools/ncbi.tar.gz the ncbi toolkit
- install it according to the README from NCBI under JCLUSTERSERVICEROOTDIRECTORY/applications/bin
- copy "blastall" and "formatdb" from "ncbi/build" into JCLUSTERSERVICEROOTDIRECTORY/applications/bin
- copy the whole "data" directory from "ncbi/data" to the location JCLUSTERSERVICEROOTDIRECTORY/applications/data
9.5.3. CLUSTALW
- download from ftp://ftp.ebi.ac.uk/pub/software/unix/clustalw/clustalw1.83.UNIX.tar.gz
- install it according to the instructions from EBI under JCLUSTERSERVICEROOTDIRECTORY/applications/bin/
- copy clustalw into JCLUSTERSERVICEROOTDIRECTORY/applications/bin
9.6. Adapt the JCLUSTERSERVICEROOTDIRECTORY/config/clusterservice-definition.xml for your local application installation
9.6.1. Change the REPOSITORYMANAGER service definition:
program-path="JCLUSTERSERVICEROOTDIRECTORY/applications/bin/repomanager"
e.g.
program-path="/home/maspectras/maspectrasjclusterservice/applications/bin/repomanager"
and the path parameter:
<parameter position="0" mandatory="true" name="repository" switch="-r">
<parameter-value isdefault="true">JCLUSTERSERVICEROOTDIRECTORY/repository</parameter-value>
</parameter>
e.g.
<parameter position="0" mandatory="true" name="repository" switch="-r">
<parameter-value isdefault="true">/home/maspectras/maspectrasjclusterservice/repository</parameter-value>
</parameter>
9.6.2. Change the NCBI-BLAST/NCBI-FORMATDB service definition:
program-path="JCLUSTERSERVICEROOTDIRECTORY/applications/bin/blastall"
e.g.
program-path="/home/maspectras/maspectrasjclusterservice/applications/bin/blastall"
and
program-path="JCLUSTERSERVICEROOTDIRECTORY/applications/bin/formatdb"
e.g.
program-path="/home/maspectras/maspectrasjclusterservice/applications/bin/formatdb"
9.6.3. Change the CLUSTALW service definition:
program-path="JCLUSTERSERVICEROOTDIRECTORY/applications/bin/clustalw"
e.g.
program-path="/home/maspectras/maspectrasjclusterservice/applications/bin/clustalw"
9.6.4. Change the MASSQUANTIFICATION service definition:
program-path="JCLUSTERSERVICEROOTDIRECTORY/applications/bin/msQuantificationCmd"
e.g.
program-path="/home/maspectras/maspectrasjclusterservice/applications/bin/msQuantificationCmd"
9.6.5. The default installation does not require any queueingsystem or cluster and therefore every service
definition is sent to the built-in local java jms queueing system. If you want to run CLUSTALW,
BLAST- and FORMATDB- jobs on a distributed cluster just adapt the configuration parameters described in
3.2 and change in the service definition the parameter queuetype="jms" to queuetype="sge" or
queuetype="pbs".
In such case the JCLUSTERSERVICEROOTDIRECTORY must be shared (rw) under the same path over all
cluster nodes.
9.7. Adapt on the Maspectras side the paths in DATAROOTDIRECTORY/analyses/partitioning/pipeline.xml
<MCLProgramDirectory>/home/maspectras/mcl</MCLProgramDirectory>
<RemoteBlastDatabaseDirectory>JCLUSTERSERVICEROOTDIRECTORY/tmp/BlastDB</RemoteBlastDatabaseDirectory>
<RemoteBlastExecutableDirectory>JCLUSTERSERVICEROOTDIRECTORY/applications/bin</RemoteBlastExecutableDirectory>
<RemoteInputDirectory>JCLUSTERSERVICEROOTDIRECTORY/tmp/Input</RemoteInputDirectory>
<RemoteResultDirectory>JCLUSTERSERVICEROOTDIRECTORY/tmp/Output</RemoteResultDirectory>
e.g.
<MCLProgramDirectory>/home/maspectras/mcl</MCLProgramDirectory>
<RemoteBlastDatabaseDirectory>/home/maspectras/maspectrasjclusterservice/tmp/BlastDB</RemoteBlastDatabaseDirectory>
<RemoteBlastExecutableDirectory>/home/maspectras/maspectrasjclusterservice/applications/bin</RemoteBlastExecutableDirectory>
<RemoteInputDirectory>/home/maspectras/maspectrasjclusterservice/tmp/Input</RemoteInputDirectory>
<RemoteResultDirectory>/home/maspectras/maspectrasjclusterservice/tmp/Output</RemoteResultDirectory>
and the connection parameter ServerURL and ServerURLPlusUser in DATAROOTDIRECTORY/analyses/partitioning/cluster.properties.
Default connection settings presume that the jclusterservice is running on the same host as maspectras on http. If its
runnning on a different machine replace the http://username:password@localhost:21080 by the proper host and port
parameters in DATAROOTDIRECTORY/analyses/partitioning/cluster.properties . If possible secure the connection via SSL
and replace in ServerURL and ServerURLPlusUser the http by https.
9.8. Start the jclusterservice under $JBOSS_HOME/bin with run-jclusterservice.sh
The jclusterservice can run on a different execution host or a cluster system. In
that way the applications queued and executed there are not influencing the performance
of the webinterface.
10. Installation of the markov cluster
10.1. Download the MCL package from http://micans.org/mcl/ and install it. mcl 1.005, 05-118 has been
tested with Maspectras
Maspectras requires the programs clmformat, mcl, and mcxassemble.
10.2. Specify the path to the MCL package in the pipeline.xml file located in the
$DATAROOTDIRECTORY/analyses/partitioning
directory:
e.g.: /home/maspectras/mcl
Make sure that the maspectras user has access to /home/maspectras/mcl and to $DATAROOTDIRECTORY
11. Get access to the AAS
11.1. Open your Web browser and type http://localhost:17080/UsermanagementWeb
11.2. Login using username: admin password: 12345
(change the password afterwards, since this is the ADMINISTRATOR account for the whole AAS)
11.3. Logout
11.4. Login using username: maspectrasadmin password: 12345
(change the password afterwards, since this is the ADMINISTRATOR account only for maspectras)
12. Get access to Maspectras
12.1. Open your Web browser and type http://localhost:8080/maspectras
13. Test Maspectras
13.1. Open a new browser and type https://maspectras.genome.tugraz.at
13.2. Login as guest user (username: guest ; password: guest)
13.3. In the left menu click "Management" -> "File Upload"
13.4. Click on "Query", type "Name" "=" "ICPL_Protmix_1lizu1he_A_c1_ms2_Mascot" and download the Mascot file (for help see user manual 1.2.1 and 1.2.3)
13.5. Click on "Query", type "Name" "=" "ICPL_Protmix_1lizu1he_A_c1_Centroid" and download the mzXML file containing the raw data (check if the file suffix is mzXML and not mzXML.xml, if not change the suffix to mzXML)
13.6. Logout as guest user and continue working at http://localhost:8080/maspectras
13.7. Upload and initialize the database doc/testdatabase/karl_04090683392.fa with the parsing rules of doc/parsingRules/giDatabase_parserules like in the user manual described at chapter 2.1
13.8. Upload the Mascot file (File Type = mascot) and the mzXML file (File Type = Raw-File) like in the user manual described at chapter 2.2
13.9. When you started the upload of the mzXML file the file is translated into a file format more convenient for calculation. Check the status of this translation at Uploadstatus (user manual 2.3). Wait until "TRANSLATING" is replaced by "TRANSLATION FINISHED"
13.10. Create a new Massspecexperiment (for help see user manual 6.1) and assign the mzXML as RawFile
13.11. Start the parsing of the Mascot file (for help see user manual 6.2)
13.12. When the parsing is finished check your results (for help see user manual 7 and 7.1). In the clustered protein view must be 16 protein(-clusters), then the clustering worked.
14. Installation of Java3D for the client machine
To use the real 3D view Java3D must be installed on the client machine. To download Java3D visit the web-site http://java.sun.com/products/java-media/3D/download.html. It is recommended to increase the memory size for the applet, for more information visit the Maspectras user manual 7.5.1.