Fuse: Multiple Network Alignment via Data Fusion
Vladimir Gligorijevic, Noel Malod-Dognin and Natasa Przulj
Corresponding author: Prof. Natasa Przulj, e-mail: natasa [AT] cs.ucl.ac.uk
Fuse is a novel global multiple network aligner. First, it computes novel similarity
scores between proteins by fusing sequence similarities and network wiring patterns
over all proteins in all PPI networks being aligned, using non-negative matrix tri-factorization (NMTF).
Second, it construct a one-to-one global multiple network alignment by using an
approximate maximum weight k-partite matching solver.
How to run Fuse
Data-fusion step
To fuse PPI networks, you need:- MATLAB R2013a+
- List of all PPI network files in edgelist format
- An input file contaning the sequence similarity between the proteins (e.g., 1-eval)
- Number of NMTF iterations
- List of rank parameters
- Gamma parameter
Example of command line (in MATLAB console):
run_nmtf({'./Nets/CElegans.edgelist', './Nets/DMelanogaster.edgelist', './Nets/HSapiens.edgelist', './Nets/MMusculus.edgelist', './Nets/SCerevisiae.edgelist'}, 'sequence_scores.lst', [80, 90, 80, 70, 50], 1000, 0.01, 'nmft_scores.lst")
Data-fusion step produces an output file "nmft_scores.lst", which is an input file for the alignment step.
Alignment step
To align PPI networks with Fuse, you need:- The PPI network files, in leda format
- An input network list file, which contains the full path to each network (one per line), e.g.:
- A file contaning the sequence similarity between the proteins (e.g., 1-eval)
- A file contaning the NMTF-based similarity between the proteins
./Nets/DMelanogaster.gw
./Nets/HSapiens.gw
./Nets/MMusculus.gw
./Nets/SCerevisiae.gw
Example of command line: ``./FUSE.exe -n network_list.txt -s sequence_scores.lst -t nmft_scores.lst -o an_output_file''
You can alternatively run ``./FUSE.exe -h'' to get the complete list of parameters.
