NAME

ElMo-Comp (GA) - Elementary Mode Computation using Global Arrays Toolkit

Elementary flux mode can be used on the top of Global Arrays Toolkit layer to support both memory and time scalability. Following are the instruction for installation of Global Arrays Toolkit and the ElMo-Tools version of the software on the top of it. Result should be a program under the name elmocompga

GLOBAL ARRAYS COMPILATION AND INSTALLATION

To compile and install download the Global Arrays Toolkit (tested on version 5.1). Following configuration will require previous installation of MPI library (tested using OpenMPI, Intel MPI and Platform MPI).

./configure --prefix=/path/to/ga-5-1/ --enable-cxx --with-mpi --with-blas=no --with-lapack=no --with-scalapack=no

Configuration excludes linking with the linear algebra libraries BLAS/Lapack/Scalapack, requires binding with C++ and specifies explicitly the path where to install Global Array Toolkit. Following, compilation, installation and testing may be performed as

$ make

$ make install

$ make check
Before moving on to the next step, you may want to set the PATH and LD_LIBRARY_PATH variables in the appropriate shell configuration file. In addition, by running command

$ make flags

from the directory where Global Arrays was installed it will print out the string that should be appended in the configure file to the LIB variable. This will assure correct compilation and linking of the ElMo-Tools GA software.

ELMO-COMP INSTALLATION

Configuration of the ElMo-Tools can be performed using the following command, where /path/to/ga-5-1 is a path where you installed Global Arrays Toolkit.
$ ./configure CXX=mpiCC --with-mpi=yes --with-ga=/path/to/ga-5-1/


Further, after configuration, compile the files into executable as:

$ make elmocompga

To run the software, standard MPI program mpirun or mpiexec can be used

$ mpirun -np 4 ./elmocompga -f data/iCT_Ecoli_59.txt

GA-based version of the software only outputs elementary flux modes in the binary bit-compressed format, as its primary intention is to be run for larger metabolic networks where the output file with fully expanded floating point numbers would be irrationally big. If nump processors are used in execution then the output will be saved in chunks with extensions *.proc_id where proc_id=0...num-1
If the option to give the output file name is given as '-b ecoli59.bin':

$ mpirun -np 4 ./elmocompga -f data/iCT_Ecoli_59.txt -b ecoli59.bin

the output will be saved into files ecoli59.bin.0, ecoli59.bin.1, ecoli59.bin.2 and ecoli59.bin.3 These output files may further be merged into a single binary output file using the Perl script merge_bin.pl provided in browse folder of the provided software installation.

$ merge_bin.pl --input binfile.list --output out.bin

The call to merge_bin.pl will accept as input a small file with paths to binary files which should be merged while the output option specifies the name of the output file.

Last change: August 2012