Random Walk with Resistance (RWS)
**The supplementary material for Bioinformatics manuscript**
The matlab code can be downloaded here.
Our network reconstruct strategy has three steps:
- Step 1, use the RWS algorithm to build the full topological profiles matrix from the original network.
- Step 2, get the correlation between each pair of the nodes.
- Step 3, rebuild the network by keep the same edges as the original network.
If you want to "one click" improve the network (step1,2,3), use
ReNet.m .
If you want to decide how to add/remove the edges by yourself or want to
know the similarities between each pair of the nodes (step1,2), use
CorrByRWS.m .
If you want to do some further research on the topological profiles of
the network (step1), use RWS.m .
About the Data and Results
The core PPI network we used in the paper can be downloaded here, which is from "N.J. Krogan, G. Cagney, H. Yu, et al. Global landscape of protein complexes in the yeast saccharomyces cerevisiae. Nature, 440:637–643, 2006".
We got some good results when we applied the algorithm on protein protein interaction (PPI) network. The correlation matrix and improved PPI network can be downloaded in .mat format. The protein names for the nodes are listed here.
We applied the algorithm on Human PPI network(HPRD, version 9). The correlation matrix can be downloaded in .mat format. The protein names for the nodes are listed here.
Additional Information
- More experimental results on different yeast/human PPI network
-
Two more yeast PPI network (yeast two-hybrid and protein-fragment complementation assay), and one human PPI network (HPRD, version 9) are tesed in this Bioinformatics paper.
-
RWS parameter effect, Krogan data's BP & CC evaluation, and human HPRD PPIs' topological similarity distribution are shown in the supplementary materials.
-
- More "adding/removing edges" strategies were discussed