---------------------------------
IPred
---------------------------------
IPred is a program that automatically combines the gene predictions of ab initio and evidence based
gene finders.
---------------------------------
Copyright (c) 2014,
Franziska Zickmann,
ZickmannF@rki.de, Robert Koch-Institute, Berlin, Germany
Distributed under the GNU Lesser General Public License, version 3.0.
When using IPred, please cite the following manuscript: in submission.
---------------------------------
INSTALLATION
---------------------------------
IPred was designed and tested on a linux system with the following minimum requirements for installed software:
- Python v 2.7 (http://www.python.org/) with additional packages Matplotlib and Numpy (only necessary when run
from source).
If you want to evaluate the input gene predictions and their combinations against a reference annotation, IPred
requires Cuffcompare, which is part of the Cufflinks suite (http://cufflinks.cbcb.umd.edu/).
To install IPred, download the compressed zip folder from https://sourceforge.net/projects/ipred/ and unpack the package with:
> unzip IPred.zip
This creates a folder named "IPred" in your current directory. This folder includes the GUI executable IPred.jar, the python
source IPred.py, a folder named "converter" which includes converter scripts for several gene finders, and a folder named "script"
which contains the precompiled executables for different os systems (IPred.exe for windows, IPred for linux and IPred_mac for Mac OS 10.6+).
To use the IPred GUI, java 7 (http://www.java.com/) needs to be installed on your system.
To receive the help message of IPred on the command line, type:
> python IPred/IPred.py --help
When using the GUI, choose "Console/show Readme.txt".
Note that if you run IPred with the help of the GUI, the folder "script" needs to be placed in the directory containing the
IPred.jar file.
---------------------------------
RUN IPred - EXAMPLE
---------------------------------
To run IPred from the command line, simply type:
> python IPred.py [options]
To be able to run IPred, at least the parameters "-i" and "-e" have to be set to provide text files specifying the input
prediction files that need to be merged.
You can start the IPred GUI with a double click on the file IPred.jar; however, if you want to
start the IPred GUI from the command line, type:
> java -jar IPred.jar
When using IPred with the GUI, please make sure that the "script" folder is in the same directory as the jar file.
To apply IPred on the example data set that can be downloaded from https://sourceforge.net/projects/ipred/files/Example/, please
change to the directory containing the program IPred.py. In the following example we assume that both the python file
and the "example" folder are placed in the directory "./IPred/". Since the paths specified in the ab initio and evidence
based input files need to be absolute, please adapt the paths in the example files example/input_abInitio.txt and
"example/input_evBased.txt" and in the example below accordingly for your installation.
> cd ./IPred/
> python IPred.py -i example/input_abInitio.txt -e example/input_evBased.txt -o example/out/ -p
To run IPred and simultaneously evaluate the predictions using CuffCompare, type:
> python IPred.py -a example/annotation.gtf -i example/input_abInitio.txt -e example/input_evBased.txt -o example/out/ -p -f
---------------------------------
Input files
---------------------------------
Various prediction outputs can be passed to IPred. They need to be classified either as "ab initio" or as "evidence based".
For each category, the name of the prediction method (or any other suitable name) and the location of the
prediction output file need to be specified in one input file that is passed to IPred with the "-i" or "-e" option, respectively.
Each method name and path to the output file need to be tab-separated, and if more than one method is provided, each line
contains the features of one method.
For example, if we have the methods foo1 and foo2 with their prediction files located in foo/foo1_pred.gtf and foo/foo2_pred.gtf,
respectively, the corresponding input file has the following content:
foo1 foo/foo1_pred.gtf
foo2 foo/foo2_pred.gtf
The folder "examples" provided with the IPred installation contains two examples for an ab initio and an evidence based input file
that correspond to the provided example data set.
---------------------------------
PARAMETERS OF IPred
---------------------------------
-h, --help show help message and exit
-v LOGNAME, --log LOGNAME Specify the name tag of the output files. Default: stats
-a ANNOPATH, --annotation ANNOPATH Reference annotation in gtf format.
-i ABINITIO, --abInitio ABINITIO Specify input file with ab initio predictions in gtf format. For format requirements, see Section "Input files" above.
-e EVBASED, --evBased EVBASED Specify input file with evidence based predictions in gtf format. For format requirements, see Section "Input files" above.
-b, --verifyEvBased If set, also evidence based predictions are verified with ab initio predictions.
-o OUTDIR, --out OUTDIR Output directory. Default: current directory.
-t THRESH, --threshold THRESH Threshold for required overlap of different predictions (percent of length). Default: 0.8.
-p, --prokaryote If set, the organism of interest is a prokaryote.
-f, --cuffcompare If set, perform a cuffcompare analysis of the gene predictions.
