The model simulates tumour growth in the presence of blood vessels. A special case of tumour cords is addressed.
For general description of the modelling framework based on mixture theory see
- Astanin, S., & Preziosi, L. (2008). Multiphase models of tumour growth. In Selected topics in cancer modeling (pp. 1-31). Birkhäuser Boston. DOI: 10.1007/978-0-8176-4713-1_9
The basic tumour cord model was explained in
- Astanin, S., & Tosin, A. (2007). Mathematical model of tumour cord growth along the source of nutrient. Mathematical Modelling of Natural Phenomena, 2(03), 153-177. DOI: 10.1051/mmnp:2007007 (free access)
The full model, which includes transition of the tumour to the anaerobic metabolism, was published in
- Astanin, S., & Preziosi, L. (2009). Mathematical modelling of the Warburg effect in tumour cords. Journal of theoretical biology, 258(4), 578-590. DOI: 10.1016/j.jtbi.2009.01.034
Running simulations is typically something like this:
$ cord -v -t 100.0 -d 0.5 --dump-period 10.0 2>log ; tail log
Model and method parameters are defined in cord.ini file. By default the file in the current directory is used. If it is not found, the programme looks for a file installed with the $prefix/share/cord. If neither file is found, the default values are used.
Command line options override settings given in the cord.ini file.
For more information about options see
$ cord --help
Results are saved in HDF5.0 files. OpenDX (.dx) and Gnuplot (.gp) files may be generated for visualization.
OpenDX (.dx) files may be generated from HDF files like this:
$ cord_h5todx *.h5
Gnuplot (.gp) files may be generated from HDF files like this:
$ cord_h5togp *.h5
To produce visualizations from Gnuplot (.gp) files using gnuplot:
$ cord_gp2eps -gnuplot *.gp
To produce visualizations from Gnuplot (.gp) files using Tioga:
$ cord_gp2pdf *.gp
To produce visualizations from Gnuplot (.gp) files using GRI:
$ cord_gp2eps -gri *.gp
HDF5 utils can also convert HDF data to VTK format, suitable for further processing with VTK or MayaVi. To extract packing density profile (variable phi) from the data file run
$ h5tovtk -d '/mesh/phi' file.h5
To extract oxygen distribution (variable c) run
$ h5tovtk -d '/mesh/c' file.h5
And to extract level set function (variable psi) run
$ h5tovtk -d '/mesh/psi' file.h5
This programme depends on the following libraries:
- libpopt (command line parsing, required)
- Blitz++ (internal matrix representation, required)
- BLAS (either ATLAS, or netlib's BLAS, or GotoBLAS, required)
- GNU Scientific Library (ODE solver, numerical integration, required)
- UMFPACK/libufsparse (generic SLE solver, optional, recommended)
- LAPACK (tridiagonal solver, optional)
- libhdf5 (data file format, optional)
The programme is written in C++, and was primarily developed with GNU C++ compiler version 3.4.
Currently the source is distributed together with LSolver package written by C. Badura (lsolver/ subdirectory) and iniParser by N. Devillard (iniparser/ subdirectory).
Tested in:
- Debian GNU/Linux (etch) i386
- Windows XP SP2 + Cygwin
The project uses GNU autoconf/automake build system. The typical steps to bootstrap the built system are:
$ autoheader
$ aclocal
$ automake --add-missing
$ autoconf
You may build and install it running ./configure script,
make and make install. See INSTALL file for more details.
Make sure that you have all the dependencies installed (see above).
On my Debian system I usually run `configure' as
$ ./configure --with-atlas=/usr/bin/sse2
as SSE2-optimized ATLAS libraries are installed in /usr/bin/sse2 on my system. Alternatively, you may build the package against legacy BLAS from netlib.org. I do it in my Cygwin installation this way:
$ ./configure --with-fblas
as libfblas.a is in /usr/local/lib on my system.
If you have some of your libraries installed in non-standard path, you can set LDFLAGS shell variable before running configure. For example, if I have popt library in $HOME/usr/lib64, I may run configure like this:
$ LDFLAGS="-L$HOME/usr/lib64" ./configure
Another example, if you installed blitz++ library to $HOME/usr, its headers will be in $HOME/usr/include. The compiler will be able to find them if you set CPPFLAGS variable, e.g.:
$ CPPFLAGS="-I$HOME/usr/include" \
LDFLAGS="-L$HOME/usr/lib" \
./configure
If BLAS is not provided in your system (like in Cygwin), you may build it yourself from source. ATLAS is a better option, but I BLAS is easier to start with.
-
Get BLAS (blas.tgz) from http://www.netlib.org/blas/
-
Unpack the BLAS source package and go to inside:
$ tar zxf blas.tgz $ cd BLAS -
Compile all the Fortran files:
$ g77 -c -O2 *.f -
Assemble all the object files produced into library libblas.a. You may later link to this library statically.
$ ar rvs libblas.a *.o -
Install the library into the system, e.g. copy it to /usr/local/lib
$ cp libblas.a /usr/local/lib/
You can obtain better performance with ATLAS. You may find instructions for Cygwin installation in http://www.ifp.uiuc.edu/~nakazato/tips/cygwin_atlas.html. Most Linux users are likely to be happy with the ready-to-use ATLAS packages of their distro.
The model has been developed at Politecnico di Torino, Italy. Research and development was funded within the 5th Marie Curie Research Training Network.
The code is distributed under the terms of the GNU General Public License version 2 or later. See COPYING for details. iniParser and LSolver distributed with the code are not covered by this license.
- Sergey Astanin
<sergey dot astanin at polito dot it> - Luigi Preziosi
<luigi dot preziosi at polito dot it> - Andrea Tosin
<andrea dot tosin at polito dot it>