| Home |
Setup the VSL2 Protein Disorder Predictor on Your Own Machine
Kang Peng, Center for Information Science and Technology, Temple University
-----------------------------------------------------------------------------------
ATTENTION: PLEASE READ THE FOLLOWING TERMS AND CONDITIONS CAREFULLY BEFORE USING
THE VSL2 PREDICTOR SOFTWARE. BY USING THIS SOFTWARE, YOU AGREE TO BE BOUND BY THE
FOLLOWING TERMS AND CONDITIONS. IF YOU DO NOT ACCEPT THESE TERMS, DO
NOT USE THIS SOFTWARE.
1) The software should be used for non-commercial purpose only;
2) The software may not be redistributed in any forms, either as a standalone
package or incorporated into another software or website, without the
author's permission;
3) The software is provided "as is". There is no warranty of any kind. The
author is not liable for any damages, harms or other consequences, which may
result from the use or inability to use of the software.
-----------------------------------------------------------------------------------
1. SYSTEM REQUIREMENTS
Since the VSL2 itself is implemented in Java, it should be able to run on any
systems with Java Runtime Environment installed. However, its actual usability
depends on whether the PSI-BLAST, PHDsec and PSIPRED software can be setup on
you system.
The current VSL2 predictor has been tested on several Linux/Unix systems with
x86 architecture. Two VSL2 variants, VSL2B and VSL2P, were also tested on Windows
systems.
2. INSTALLATION
Let's assume that $VSL2DIR is the directory where VSL2 predictor will be installed.
1) Download and install the Java(TM) Runtime Environment (JRE), version 1.4.2
or higher. You may want to update the PATH environment variable to include
the JRE executable directory.
http://java.sun.com/j2se/desktopjava/jre/index.jsp
2) Uncompress the downloaded file "VSL2.tar.gz" into directory $VSL2DIR
% cp VSL2.tar.gz $VSL2DIR
% cd $VSL2DIR
% tar zxvf VSL2.tar.gz
"VSL2.jar" is the executable file, which contains 8 predictor models,
including VSL2B, VSL2P and VSL2, each using different features combinations.
Please see the manuscript and the USAGE below for more details.
NOTE: If you just want to use VSL2B, Step 1)-2) are enough. If you want to
use VSL2P, Step 5) and 6) can be omitted.
3) Download the PSI-BLAST package "blast-2.2.12-ia32-linux.tar.gz" from
ftp://ftp.ncbi.nlm.nih.gov/blast/executables/release/2.2.12/
Copy it to directory $VSL2DIR, and uncompress it there
% cp blast-2.2.12-ia32-linux.tar.gz $VSL2DIR
% cd $VSL2DIR
% tar zxvf blast-2.2.12-ia32-linux.tar.gz
% mv blast-2.2.12-ia32-linux/ blast/
% cd blast/
% mv blast/bin/* .
4) Download the UniRef100 sequence database "uniref100.fasta.gz" from
ftp://ftp.uniprot.org/pub/databases/uniprot/uniref/uniref100/
Copy it to directory $VSL2DIR/blast/, and uncompress it there
% cp uniref100.fasta.gz $VSL2DIR/blast/
% cd $VSL2DIR/blast/
% gzip -d uniref100.fasta.gz
Use the formatdb program comes with PSI-BLAST to format the UniRef100
database
% ./formatdb -i uniref100.fasta -p T -o T -n ur100
5) Download the PHDsec (v2.1) predictor from
ftp://cubic.bioc.columbia.edu/pub/rost/phd/install.pl
ftp://cubic.bioc.columbia.edu/pub/rost/phd/phd_LINUX.tar.gz
Please follow its instruction to install it into directory $VSL2DIR/phd/
6) Download the PSIPRED (v2.4) predictor from
http://bioinf.cs.ucl.ac.uk/psipred/
Please follow its instruction to install it into directory $VSL2DIR/psipred/,
make necessary changes to its main file "runpsipred" and prepare the
uniref100 database as required.
3. USAGE
The command line to run the VSL2 predictor is,
% java -jar VSL2.jar <options>
where possible options are:
-s:<sequence file> REQUIRED - Sequence should include the single-letter AA
code only; no FASTA header, no 'X', 'Y', 'Z' or other
characters
-p:<PSSM file> OPTIONAL - PSI-BLAST profile (PSSM)
-h:<PHDsec file> OPTIONAL - PHDsec prediction (*.rdbPhd file)
-i:<PSIPRED file> OPTIONAL - PSIPRED prediction (*.ss2 file)
-w:<output window> OPTIONAL - Must be an odd integer; default value is 1
The predictor program contains 8 models, each using different features
combination. Based on the options used, the program determines which model to
invoke. Following are three examples.
options used predictor invoked
----------------------------------------
-s VSL2B
-s, -p VSL2P
-s, -p, -h, -i VSL2
Other combinations are also possible. Please refer to the manuscript for details.
4. AN EXAMPLE
Assume file "testseq.fasta" contains the sequence in FASTA format, file
"testseq.flat" contains the sequence only (i.e. no FASTA header), and both files
are located in VSL2 installation directory $VSL2DIR.
1) Generating PSI-BLAST profile.
% cd $VSL2DIR/blast
% ./blastpgp -i ../testseq.fasta -d ur100 \
-h 0.0001 -e 0.0001 -j 3 -Q ../testseq.pssm > trash.txt
Where "blastpgp" is the PSI-BLAST program, "trash.txt" will contain all outputs
and is not used, and "testseq.pssm" is the file receiving the PSI-BLAST profile.
3) Make PHDsec secondary structure prediction
% cd $VSL2DIR/phd
% ./phd.pl ../testseq.fasta sec
% mv testseq.rdbPhd ../
% rm testseq.*
Where file "testseq.rdbPhd" contains the PHDsec prediction, all other output
files are discarded.
4) Make PSIPRED secondary structure prediction
% cd $VSL2DIR/psipred
% ./runpsipred ../testseq.fasta
% mv testseq.ss2 ../
% rm testseq.*
Where file "testseq.ss2" contains the PSIPRED prediction, all other output
files are discarded.
5) Finally, make the prediction
% cd $VSL2DIR
% java -jar VSL2.jar -s:testseq.flat -p:testseq.pssm \
-h:testseq.rdbPhd -i:testseq.ss2 -w:1 > testseq.pred
Where file "testseq.flat" contains the sequence without FASTA header, and the
final (numeric) prediction will be written into file "testseq.pred"
5. WEB SERVICE
The VSL2B and VSL2P predictors are also freely accessible for non-commericial
use via our web service at
http://www.ist.temple.edu/disprot/predictorVSL2.php
However, due to available computational resources, only limited number of
predictions can be provided per IP address per day.
CITATION
Peng K., Radivojac P., Vucetic S., Dunker A.K., and Obradovic Z., Length-Dependent
Prediction of Protein Intrinsic Disorder, BMC Bioinformatics 7:208, 2006.
Obradovic Z., Peng K., Vucetic S., Radivojac P., and Dunker A.K., Exploiting
Heterogeneous Sequence Properties Improves Prediction of Protein Disorder, Proteins
61(S7):176-182, 2005.