EzDSD is a software package for biologists to easily cluster biological networks using an integrated DSD (diffusion state distance) and spectral clustering approach.
The clustering method used here is the winning method (K1) developed for the DREAM challenge.
EzDSD offers the following features for network analysis:
- The ability to easily cluster groups of similar nodes to help mine important data about biological networks.
- To run the DSD on its own to retrieve either a similarity or dissimilarity matrix describing the relationships between nodes.
EzDSD also packages tools that score how significant these modules (clusters) are with respect to their association with complex traits and diseases. Of course, which traits/diseases are used is configurable.
It uses two different methods for this:
- The FuncAssociate Service which measures functional enrichment of a module.
- The PASCAL tool which scores modules using pathway analysis.
Both scoring tools have their benefits. FuncAssociate is orders of magnitude quicker to run and more easily configurable. PASCAL is the tool used in the DREAM challenge and yields more nuanced information on the quality of clusters.
The package is structured as follows:
EzDSD
| README.md
|
|---Scoring
| |
| |---FuncAssociate
| | | README.md
| | | ...
| |
| |---PASCAL
| | README.md
| | ...
|
|---Clustering
| | README.md
| | ...
The top level directory contains installation, configuration scripts, and documentation for the package as a whole. Lower level directories contain the same with the appropriate level of granularity.