dGAselID: dGAselID
In dGAselID: Genetic Algorithm with Incomplete Dominance for Feature Selection

Description Usage Arguments Value Examples

Initializes and starts the search with the genetic algorithm.

dGAselID(x, response, method = knn.cvI(k = 3, l = 2), trainTest = "LOG",
  startGenes, populationSize, iterations, noChr = 22, elitism = NA,
  ID = "ID1", pMutationChance = 0, nSMutationChance = 0,
  fSMutationChance = 0, lSDeletionChance = 0, wChrDeletionChance = 0,
  transposonChance = 0, randomAssortment = TRUE, embryonicSelection = NA,
  EveryGeneInInitialPopulation = TRUE, nnetSize = NA, nnetDecay = NA,
  rdaAlpha = NA, rdaDelta = NA, ...)

`x`	Dataset in ExpressionSet format.
`response`	Response variable
`method`	Supervised classifier for fitness evaluation. Most of the supervised classifiers in MLInterfaces are acceptable. The default is knn.cvI(k=3, l=2).
`trainTest`	Cross-validation method. The default is "LOG".
`startGenes`	Genes in the genotypes at initialization.
`populationSize`	Number of genotypes in initial population.
`iterations`	Number of iterations.
`noChr`	Number of chromosomes. The default value is 22.
`elitism`	Elite population in percentages.
`ID`	Dominance. The default value is "ID1". Use "ID2" for Incomplete Dominance.
`pMutationChance`	Chance for a Point Mutation to occur. The default value is 0.
`nSMutationChance`	Chance for a Non-sense Mutation to occur. The default value is 0.
`fSMutationChance`	Chance for a Frameshift Mutation to occur. The default value is 0.
`lSDeletionChance`	Chance for a Large Segment Deletion to occur. The default value is 0.
`wChrDeletionChance`	Chance for a Whole Chromosome Deletion to occur. The default value is 0.
`transposonChance`	Chance for a Transposon Mutation to occur. The default value is 0.
`randomAssortment`	Random Assortment of Chromosomes for recombinations. The default value is TRUE.
`embryonicSelection`	Remove chromosomes with fitness < specified value. The default value is NA.
`EveryGeneInInitialPopulation`	Request for every gene to be present in the initial population. The default value is TRUE.
`nnetSize`	for nnetI. The default value is NA.
`nnetDecay`	for nnetI. The default value is NA.
`rdaAlpha`	for rdaI. The default value is NA.
`rdaDelta`	for rdaI. The default value is NA.
`...`	Additional arguments.

The output is a list containing 5 named vectors, records of the evolution:

`DGenes`	The occurrences in selected genotypes for every gene,
`dGenes`	The occurrences in discarded genotypes for every gene,
`MaximumAccuracy`	Maximum accuracy in every generation,
`MeanAccuracy`	Average accuracy in every generation,
`MinAccuracy`	Minimum accuracy in every generation,
`BestIndividuals`	Best individual in every generation.

## Not run: 
 library(genefilter)
 library(ALL)
 data(ALL)
 bALL = ALL[, substr(ALL$BT,1,1) == "B"]
 smallALL = bALL[, bALL$mol.biol %in% c("BCR/ABL", "NEG")]
 smallALL$mol.biol = factor(smallALL$mol.biol)
 smallALL$BT = factor(smallALL$BT)
 f1 <- pOverA(0.25, log2(100))
 f2 <- function(x) (IQR(x) > 0.5)
 f3 <- ttest(smallALL$mol.biol, p=0.1)
 ff <- filterfun(f1, f2, f3)
 selectedsmallALL <- genefilter(exprs(smallALL), ff)
 smallALL = smallALL[selectedsmallALL, ]
 rm(f1)
 rm(f2)
 rm(f3)
 rm(ff)
 rm(bALL)
 sum(selectedsmallALL)

 set.seed(149)
 res<-dGAselID(smallALL, "mol.biol", trainTest=1:79, startGenes=12, populationSize=200,
               iterations=150, noChr=5, pMutationChance=0.0075, elitism=4)
 
## End(Not run)