ASGARD is a genetic algorithm designed for rule mining

Rule mining

A rule is a knowledge representation in the form:
IF cond_1 AND cond_2 ... AND cond_m THEN pred
This kind of rules contains two parts : the IF part which is called the rule condition and the THEN part which is called the rule prediction. The rule condition (C) contains a conjunction of $m$ conditions about values of predictor attributes. The rule prediction (P) contains the goal attribute.

ASGARD

Genetic Algorithm (GA) works by repeatedly modifying a population of artificial structures through the application of genetic operators. The goal is to find the best possible solution or, at least good, solutions for the problem. For this particular biological problem, we have developed a genetic algorithm named ASGARD (Adaptive Steady state Genetic Algorithm for association Rule Discovery)

INPUT FILES

Attributs.txt

This file contains the name of the attributes. There is no heading in the file. Each attribute is on a line without an extra ending line at the end of the file.

SNP1
SNP2
SNP3
Status

Goalsfile.txt

This file could contains 0, 1 or more attribute to be put in the P part.

Status

genome.txt

This file contains all the data in a specific format. Each instance is on its line, each attribute is separated by a comma.

aa,bb,ab,affected
ab,bb,aa,affected

OUTPUT FILE

results are stored in a file called RESULTS.txt. The file contains the intesting rule and the metrics.

DOWNLOAD SOFTWARE

Asgard software The release version include a makefile and the source.
General version : for all type of data db3uci4
Specialized version : specialized for linkage disequilibrium db3bio