MCODE Class Reference

List of all members.

Public Member Functions
double	computeGraphScore (const MyGraph &graph)
	Compute the MCODE-defined score of a complex.
vector< Graph >	findComplexes (const Graph &graph, double vwp, map< string, double > *nodeWeights=NULL)
	Compute and return all complexes in graph. This method is the main entry point into the MCODE class.

---

Member Function Documentation

double MCODE::computeGraphScore

(

const MyGraph &

graph

)

Compute the MCODE-defined score of a complex.

MCODE defines the score of a graph to be the product of its completeness and the number of nodes in it. But the MCODE java implementation calculates it as the ration of the number of edges in the graph to the number of nodes in the graph.

Parameters:

[in]

graph

The graph to be scored.

Returns:

The score of the graph.

vector< Graph > MCODE::findComplexes	(	const Graph &	graph,
		double	vwp,
		map< string, double > *	nodeWeights = NULL
	)

Compute and return all complexes in graph. This method is the main entry point into the MCODE class.

Parameters:

[in]	graph	the graph to be clustered
[in]	vwp	a vertex weighting fraction. The process of finding a complex starts from a seed node, say s. A new node v will be added to the complex only if the v's weight is at least (1 - vwp) times the weight of s. In other words, a lower vwp makes the algorithm more conservative when deciding to add nodes to a complex.
[in]	nodeWeights	a map from nodes to pre-calculated weights. If this argument is passed, findComplexes() multiplies each node's weight in this argument by the internally-computed MCODE node weight.

Returns:

a vector of MyGraph containing all the complexes.

Note:

vwp must lie between 0 and 1. Any other values will cause your computer to eat you.

If nodeWeights is not NULL, it must contain weights for all nodes in the graph. Any missing nodes will automatically get a weight of 0.

Warning:

The way the method deals with nodeWeights may change in the future. For instance, one possibility is to simply bypass the internal node weight computation when nodeWeights is not NULL.

---

The documentation for this class was generated from the following files:

• /home/poirel/src/c++/biorithm/libgraph/mcode.h
• /home/poirel/src/c++/biorithm/libgraph/mcode.C