EgoNet is implemented by Python and it is designed to detecting disease related subnetwork from a large biological network (PPI, metabolic network) combined with gene expression data.
Python version 2.7 or later (http://www.python.org/)
Python packages:
- Biopython (http://biopython.org/wiki/Main_Page)
- NetworkX (http://networkx.github.io/)
- SciPy (http://www.scipy.org/)
- Numpy (http://www.numpy.org/)
- scikit-learn (http://scikit-learn.org/stable/)
python egonet.py -n <network_file> -g <gene_matrix_file> -o <output_file> [opts]
-m <int> :method of classification or regression (default: class)
-t <float> :percentage of top selected gene for searching (default: 1, no sort)
-s <float> :score cutoff for printing selected subnetwork (default: 0.6)
-f <pickle> :saved subnetwork python object used for visualization (default: subnetwork.py)
-r <txt> :saved gene list ranked by two measuring methods (default: gene_rank.txt)
-h :produce this menu
python egonet.py -n sample_data/input/network.adjlist -g sample_data/input/gene_expression.txt -o TNBC.txt -f svm_net.pk
Example data are provide in the directory sample_data/
- network file is the adjacency list format
- gene matrix file starts with gene name or entrize id as first column and expression values for other columns, the last row starts with "outcome" and labels for each sample
- the output files contain the ranked subnetwork by predicting accuracy and ranked genes names using M-value.
M = msi
where m is total number of subnetwork contained gene, s is the score of each subnetwork and i is the importance of gene.
Selected subnetwork can be plotted using as followed:
python script/drawnet.py mark_gene diff_gene network_obj gene_matrix_file node
python script/drawnet.py sample_data/visualization/breastcancer.gene sample_data/visualizaiont/diffexpress.gene svm_net.pk sample_data/input/gene_expression.txt 675
Questions, suggestions, comments, etc?
Author: Rendong Yang
Send email to cauyrd@gmail.com
Yang, Rendong, et al. "EgoNet: identification of human disease ego-network modules." BMC genomics 15.1 (2014): 314.