We offer the JAVA-package of "Long Short-Term Memory" for Protein
classification (jLSTM_protein).

:: License

LICENSE, WARRANTY, AND LIABILITY
This programm is freely available under the GNU General Public License
(GPL).
You should have received a copy of the GNU General Public License along
with this program; if not, write to the Free Software Foundation, Inc., 51
Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.

:: Requirements

A Java SDK and Apache Ant is required to build jLSTM. You can get the latest
Java SDK for free at

http://java.sun.com/

and Apache Ant here

http://ant.apache.org/

jLSTM is tested with Java 1.6 and Ant 1.6.5 and Ant 1.7.1 other versions of
Ant should do as well.


:: Building jLSTM with ant

type

ant

in the extracted jLSTM directory.

this builds the jlstm.jar

:: Running jLSTM with the included example

type

ant run

in the extracted jLSTM directory

this runs the included dataset on the a.1.1 SCOP 1.67 superfamily with
the a.1.1.2 family in the positive test set.

Without ant jLSTM is started by

(Unix environment)

java -classpath "lib/biojava.jar:lib/bytecode.jar:jlstm.jar" at.jku.bioinf.jlstm.JLSTM -p JLSTM.properties -w wm.mat -t 4

(Windows environment)

java -classpath "lib/biojava.jar;lib/bytecode.jar;jlstm.jar" at.jku.bioinf.jlstm.JLSTM -p JLSTM.properties -w wm.mat -t 4

This starts the jLSTM with four threads (-t 4).

:: Command line arguments

-w wm.mat :

Name of the weight matrix which is written or read to/from the files system

-p JLSTM.properties :

Name of the properties file for configuration. See README.parameter.

-t 4 :

Number of threads

-lw (optional excludes -test) :

Load weight matrix from file system and continue training

-test (optional excludes -lw) :

Load weight matrix from file system and make just one test run

:: Important for Remote Homology experiments

For Remote Homology experiments it's necessary to extend the positive
training sequences with PSI-BLAST. See

http://www.bioinf.jku.at/software/LSTM_protein/README.extend

and

http://www.bioinf.jku.at/software/LSTM_protein/extendSequences.tar.gz


:: Citation

Please cite:

Sepp Hochreiter, Martin Heusel, and Klaus Obermayer.
"Fast Model-based Protein Homology Detection without Alignment."
Bioinformatics 2007; doi: 10.1093/bioinformatics/btm247.

04/03/2009 mhe