Update Aug 2016:

http://abi.inf.uni-tuebingen.de/Services/MultiLoc2/multiloc2_download
indicates this was released under GPL.

I want to add some code so that the tmpfile_path can be controlled by
a config file, instead of by running sed on python source files. This
makes it easier to run multiple instances at the same time.

default behavior is unchanged. ie running
python configureML2.py
still edits the python source files with sed.
However if the file 'multiloc2.ini' is found in the current directory
or the environment variable MULTILOC2_INI exists and points to a file
then that file will be used to override tmpfile_path.

The format of the file should be:

[multiloc2]
tmpfile_path=/path/to/new/tmpdir







Introduction:
=============

MultiLoc2 is a bioinformatics tool for predicting subcellular localizations of eukaryotic proteins.
It's main purpose are large-scale annotations of protein sequence data. 
The prediction system is implemented in python and trained using support vector machines. 
MultiLoc2 combines the output of several specialized subpredictors: 
- SVMTarget: for recognizing N-terminal targeting sequences (SPs, MTPs, CTPs) 
- SVMSA: for recognizing signal anchors
- SVMaac: for analyzing the overall amino acid composition
- MotifSearch: for recognizing relevant sequence motifs like NLSs, KDELs, SKLs, or DNA binding domains
- GOLoc: for analyzing Gene Ontology terms derived from the protein sequence
- PhyloLoc: for analyzing the phylogenetic profiles of the query sequences based on 78 genomes

More information about the system architecture, training procedure and performance evaluation
can be found in:

Blum,T., Briesemeister,S. and Kohlbacher,O.
MultiLoc2: integrating phylogeny and Gene Ontology terms improves subcellular protein localization prediction
BMC Bioinformatics, 2009

MultiLoc2 is available at 
http://www-bs.informatik.uni-tuebingen.de/Services/MultiLoc2/

Requirements:
==========
MultiLoc2 needs a Unix environment and uses several external tools.
Before MultiLoc2 can be run, the Libsvm and Blast packages must be installed and available
in your local $PATH variable. Furthermore, it is recommended but not required that the InterPro package
is also installed.

Libsvm is available at
http://www.csie.ntu.edu.tw/~cjlin/libsvm

Blast is available at
http://www.ncbi.nlm.nih.gov/BLAST/download.shtml

InterPro is available at
http://www.ebi.ac.uk/interpro

Installation:
=========

To install MultiLoc2 just execute
python configureML2.py

The script sets all paths and checks if the required packages are installed.
The MultiLoc2 package contains three subdirectories:

src : all python scripts that build up the prediction system
data : the required data in the form of trained Libsvm models and NCBI genome data
tmp : for temporary files created and removed during program execution

Usage:
=====

Run MultiLoc2 without parameters to see how it can be used by the command
python src/multiloc2_prediction.py

The following list of required and optional parameters can be used to configure MultiLoc2:

-fasta=<fasta file>
	required: MultiLoc2 accepts as input protein sequence data in fasta format.
	
-origin=<animal|plant|fungal>
	required: Depending on the specified origin, different sets of eukaryotic subcellular localizations
	can be predicted.

-result=<result file>
	required: The prediction results are stored in a file given by this parameter.

[-predictor=<LowRes|HighRes>]
	optional: MultiLoc2 is available in two versions since it was trained on two different training data sets.
	The LowRes version predicts up to five localizations and the HighRes up to 10. The HighRes version 
	is used as default predictor. 

[-output=<simple|advanced>]
	optional: This parameter defines how much result information is presented. The simple output consists
	of the final probability scores for each predictable localization and is used as default output.
	The advanced output contains also detailed results for each subpredictor.

[[-go=<go file>] ... ]
	optional: The GOLoc subpredictor requires Gene Ontology (GO) terms as input.
	Several files containing GO data can be passed via this parameter. 
	The lines in the files must be in the following format:
	<protein fasta id><at least one space or tab><GO data>
	where <GO data> is an annotating text which is scanned for GO terms in the format GO:[0-9][0-9][0-9][0-9][0-9]
	
	The protein ids in the GO data files must also be present in the input sequence fasta file. 
	GO data for each protein can be distributed over several lines or several files. 
	If no file containing GO data is passed, MultiLoc2 skips the GOLoc subpredictor and replaces its output
	by default values. It is recommended to use the program iprscan from the InterPro package to create GO data.

MultiLoc2 and InterPro:
=======================

MultiLoc2 and the program iprscan from the InterPro package can be combined using the script 
run_multiloc2_with_iprscan. The script runs at first ipscan using a passed fasta file as input
and creates the file interpro.out which in turn is passed via the -go param to multiloc2_prediction.py.

To run the script type
./run_multiloc2_with_iprscan <fasta file> <origin> <result file>

You are free to change the script according to your wishes.

Examples:
=======

The following command runs the animal version of MultiLoc2(-HighRes) using test.fasta and test.go as input
and stores the prediction results in test.res:
python src/multiloc2_prediction.py -fasta=test.fasta -origin=animal -result=test.res -go=test.go

The next command runs iprscan at first using test.fasta as input. After this, the animal version of MultiLoc2(-HighRes)
is executed using again test.fasta as input and test.res as result file. This time, the GO output of iprscan
is automatically passed to MultiLoc2 via the -go parameter:
./run_multiloc2_with_iprscan test.fasta animal test.res