The Python [Powergraph analysis](https://en.wikipedia.org/wiki/Power_graph_analysis) tool,
based on [Answer Set Programming](https://en.wikipedia.org/wiki/Answer_set_programming) solving and formal concept analysis.
More technical documentation about PowerGrASP can be found in the [documentation file](https://github.com/Aluriak/PowerGrASP/blob/master/doc/documentation.mkd)
PowerGrASP results, compared to the existing BIOTEC tool, can be found in the [results file](https://github.com/Aluriak/PowerGrASP/blob/master/doc/results.mkd)
## Installation & Requirements
pip install powergrasp
Just be sure that, in the library installation directory, for [pyasp](https://github.com/sthiele/pyasp)
you have the binaries *gringo4* and *clasp* installed.
In case of a system-wide installation, these binaries are generally correctly installed and can be found in */usr/lib/python3.5/site-packages/pyasp/bin/*.
The binaries, if not properly installed by pip, can be found on the sourceforge website:
- [gringo4](http://sourceforge.net/projects/potassco/files/gringo/4.5.3/) in version 4.5.3, that probably needs to be renamed from *gringo* to *gringo4*.
- [clasp](http://sourceforge.net/projects/potassco/files/clasp/3.1.3/) in version 3.1.3.
A magical workaround is to use the *--no-cache-dir* option of pip for to the binaries properly installed, even in local or with a virtualenv:
pip install powergrasp --no-cache-dir
## Basic use
PowerGrASP can be used as a script:
python3 -m powergrasp --graph-data=human_proteom.lp --output-file=for_cytoscape.bbl
Or can be embedded in any python program:
## Help & Details
### General overview
The compression is configurable through command line arguments or compress function parameters.
Used ASP source code can be changed, interactive mode can be set,… Please look at help and docstring:
# in terminal
python3 -m powergrasp --help
# in python
# or, in the PowerGrASP git directory, with make
NB: Create another output or input format support is possible by implement a new Converter class (see *powergrasp/converter/*).
#### Input file
PowerGrASP doesn't generate logging for pleasure : it actually perform a treatment on input data, if provided.
The supported input file formats are currently :
- ASP: atoms edge/2, with edge(X,Y) describing a link between nodes X and Y.
- SBML: a regular SBML file, when species and reactions will be treated as nodes.
- GML: a regular Graph Modeling Language file, readable by networkx python module.
Other formats will be supported in the future.
##### Bubble formatted file
The output of PowerGrASP is a *Bubble* formatted file. This file can be used to get a visualization of the compressed graph.
The format *Bubble* is an endemic format designed by Royer et Al for power graph describing.
Bubble files can be used in several ways:
Print a power graph through Cytoscape is made possible by the [CyOog](http://www.biotec.tu-dresden.de/research/schroeder/powergraphs/) plugin,
which handle the format *Bubble*.
Cytoscape, in order to using the CyOog plugin, must be in version __[2.x](http://www.cytoscape.org/download_old_versions.html)__
##### Oog Command line tool
BIOTEC team also released a [command line tool](http://www.biotec.tu-dresden.de/research/schroeder/powergraphs/download-command-line-tool.html) for power graph analysis.
This tool allow to print a bubble file without Cytoscape usage, with something like:
java -jar Oog.jar -inputfiles=path/to/bubble.bbl -img -f=png
Add *&>/dev/null* to prevent any logging output.
### Standard output management
By default, PowerGrASP generates lots of outputs in stdout, essentially for debugging and compression tracking.
With the option *loglevel*, its possible to control this behavior:
python3 -m powergrasp --graph-data=tests/proteome_yeast_2.lp --loglevel=warning
This will block all outputs with a strictly lesser priority than warning.
Available levels comes from logging API:
log level | PowerGrASP
critical | totally silencious
error | very rarely disturbing
warning | rarely disturbing
info | trackable
debug | trackable with __high__ verbosity
notset | kraken released
Please note that some options (notabily *count-models* and *count-cc*) are completely independant of this logging management, as they work with the standard output.
The compression compute some statistics about itself, and generate the final results
at the end of the compression in the standard output.
With some arguments, you can also show a colored graphic :
python3 -m powergrasp --graph-data=tests/proteome_yeast_2.lp --stats-file=data/statistics.csv --plot-stats
Instead of show it, powergrasp can save it in png (note that the *--plot-stats* flag is not necessary when *plot-file* option is given):
python3 -m powergrasp --graph-data=tests/proteome_yeast_2.lp --stats-file=data/statistics.csv --plot-file=data/statistics.png
### Answer Set Programming
ASP is a declarative and logic language, designed for the treatment of combinatorial problems (like graph compression).
The implementation used in this project is the [*Potsdam Answer Set Solving Collection*](http://potassco.sourceforge.net/index.html)
All ASP source codes necessary for the PowerGrASP program can be found in *powergrasp/ASPsources/* directory.
### Interests & References
The [Power Graph approach for graph compression](https://en.wikipedia.org/wiki/Power_graph_analysis) allows a lossless compression with an emphasis on biological meaning.
In fact, formal concepts used by Power Graph analysis have a sens in biology, especially in the case of proteomes.
All graphs can be compressed through Power Graph, and will be more readable once compressed,
but interactomes, at least, also gain in interpretability.
The main inspiration of PowerGrASP : PowerGraph Analysis:
Loïc Royer, Matthias Reimann, Bill Andreopoulos, and Michael Schroeder.
Unraveling Protein Networks with Power Graph Analysis.
PLoS Comput Biol, 4(7):e1000108, July 2008.
Usage of the PowerGraph Analysis:
Loic Royer, Matthias Reimann, A. Francis Stewart, and Michael Schroeder.
Network Compression as a Quality Measure for Protein Interaction Networks.
PLoS ONE, 7(6):e35729, June 2012.
Yun Zhang, Charles A Phillips, Gary L Rogers, Erich J Baker, Elissa J Chesler, and Michael A Langston.
On finding bicliques in bipartite graphs: a novel algorithm and
its application to the integration of diverse biological data types.
BMC Bioinformatics, 15(1):110, 2014.
ASP through Potassco implementation:
M. Gebser, R. Kaminski, B. Kaufmann, M. Ostrowski, T. Schaub, and M. Schneider.
Potassco: The Potsdam answer set solving collection.
AI Communications, 24(2):107–124, 2011.
TODO: Brief introduction on what you do with files - including link to relevant help section.