GPASInteractions: Execute the GPAS algorithm for feature selection

Description

Identification of interesting (high order) SNP interactions. The algorithm works on categorical data with binary response and delivers multi-valued logic expressions in disjunctive normal form typically explaining subsets of the data and an interaction tree containing interesting interactions.

Usage

GPASInteractions(resp, preds, runs = 1, generations = 10000,  savegraph = "interactions.dot",occurences=10,ratio=0.1)

Arguments

resp

Vector with the response variables

preds

Matrix or data frame with the predictors

runs

Number of independent runs of GPAS

generations

Number of generations after which the algorithm will be stopped

savegraph

Name of the file the resulting GraphViz graph will be saved to

occurences

The minimum number of times an interaction has to occur to be included in the graph

ratio

The minimal ratio a single literal has to occur in relation to his ancestor in the interaction graph to be included in the interaction graph

Value

run: The run the individual was found in
generation: The generation the individual was created in
objective value 1: Sum of correctly predicted cases and controls
objective value 2: Correctly predicted controls
objective value 3: Length of the individual
individual: A string representation of the individual

References

R. Nunkesser, T. Bernholt, H. Schwender, K. Ickstadt, and I. Wegener (2007). Detecting High-Order Interactions of Single Nucleotide Polymorphisms Using Genetic Programming. Bioinformatics, 23, 3280-3288.

GraphViz: http://www.graphviz.org/

Examples

Run this code

# load example data
data(data.logicfs)

# execute GPAS to search for interesting interactions
GPASInteractions(cl.logicfs,data.logicfs,runs=1,generations=1000)