++ed by:
1 non-PAUSE user
and 1 contributors

# NAME

Algorithm::MCL - perl module implementing Markov Cluster Algorithm using PDL

version 0.004

# SYNOPSIS

``````    use Algorithm::MCL;

my \$obj1 = new MyClass;
my \$ref2 = {};
my \$ref3 = \"abc";
my \$ref4 = \\$val1;
my \$ref5 = [];

my \$mcl1 = Algorithm::MCL->new();

# create graph by adding edges

# run MCL algorithm on created graph
my \$clusters1 = \$mcl1->run();

# get clusters
foreach my \$cluster ( @\$clusters1 ) {
print "Cluster size: ". scalar @\$cluster. "\n";
}

####################################

my \$val1 = \"aaa";
my \$val2 = \"bbb";
my \$val3 = \"ccc";
my \$val4 = \"ddd";
my \$val5 = \"eee";

my \$mcl2 = Algorithm::MCL->new();

my \$clusters2 = \$mcl2->run();

foreach my \$cluster ( @\$clusters2 ) {
print "Found Cluster\n";
foreach my \$vertex ( @\$cluster ) {
print "  Cluster element: \$\$vertex \n";
}
}``````

# DESCRIPTION

This module is perl implementation of Markov Cluster Algorithm (MCL) based on Perl Data Language (PDL).

MCL is algorithm of finding clusters of vertices in graph. More information about MCL can be found at http://micans.org/mcl/. There is also perl script implementing MCL - minimcl http://www.micans.org/mcl/scripts/minimcl.

This module try to solve two problems:

• easy integration MCL in perl scripts and modules. Algorithm::MCL accept references as input and every reference will be found later in some cluster.

• performance and scale. Algorithm::MCL use Perl Data Language for most of its processing and should run very fast on very big clusters. Main Algorithm::MCL procedures are written with "pdlpp".

# METHODS

## new()

create new Algorithm::MCL object that accumulate graph edges and process data.

add new edge to graph. first two parameters are reference to vertex objects. third parameter is "connection strength measurement" between vetices. "connection strength measurement" should be number between 0 and 1, higher number means stronger connectivity. if "connection strength measurement" is not defined it set to 1.

## run()

apply MCL algorithm on graph. return reference to array that every element is reference to cluser array.

# AUTHOR

Pinkhas Nisanov <pinkhas@nisanov.com>