NGCluster is an experiment with clustering gene expression data using neighborhood graphs.
Ben Saylor and Londen Johnson
Dr. Rahul Singh, Professor, Computer Science, SFSU
CSC 857 - Bioinformatics Computing, Fall 2014
San Francisco State University
The easiest way to satisfy all of the dependencies is to install the Anaconda Python distribution.
- Python 3 (version 3.4.0 has been tested)
- LLVM (version 3.3 has been tested)
- See the llvmpy web site for installation instructions
- NumPy (version 1.9.1 has been tested)
- SciPy library (version 0.14.0 has been tested)
- Matplotlib (version 1.4.2 has been tested)
- If you encounter a TypeError when installing Matplotlib, the workaround described here may apply.
- llvmpy (version 0.12.7 has been tested)
- Numba (version 0.15.1 has been tested)
All but the first two packages are installable from PyPI and can be installed
using pip install -r requirements.txt.
NGCluster includes a number of configurations of graph and clustering
parameters. Running a configuration produces a set of output files in a
directory named output/configuration-key, where configuration-key is the
short name for one of the configurations listed by python run.py. Running the
program also compiles summary output for the listed and all previously run
configurations in output/compiled-results.csv.
To list available configurations:
python run.py
To run a set of configurations:
python run.py configuration-key-1 [configuration-key-2 ...]
To run all configurations:
python run.py all
To run no configurations but produce the compiled output file based on existing output:
python run.py compile