EpiTag: Selection of TagSNPs
In MathieuEmily/EpiTag: TagSNP selection for epistasis

Description Usage Arguments Details Value References See Also Examples

EpiTag allows the selection of tag SNPs in 1 or 2 genomic regions. The selection aims at optimizing the power to detect epistasis between SNPs.

1	EpiTag(Region1=NULL,Region2=NULL,mat.MI=NULL,threshold=0.8,allelic=FALSE)

`Region1`	A nxp1 `matrix` or `data.frame` containing allele or genotype of the p1 SNPs for the n individuals.
`Region2`	A nxp2 `matrix` or `data.frame` containing allele or genotype of the p2 SNPs for the n individuals. Region 1 and Region 2 must have the same number of rows.
`mat.MI`	An object of the class `MIMatrix` where the `mat.MI` element is that contains a p1xp2xp1xp2 `array`. For example, an output of the `getMatMI` function.
`threshold`	Threshold applied to the Mutual Information to create the successive bins of SNP pairs.
`allelic`	A flag indicating whether data are allelic `allelic=TRUE` or genotypic `allelic=FALSE`.

SMILE is an implementation of the statistical procedure proposed in Friguet and Emily (2013). The EpiTag method is based on the computation of the "matrix" of mutual information between all pairs of SNP pairs within a single genomic region and between 2 genomic regions. Mutual information can be computed before the selection, by using the argument mat.MI or during the selection process if mat.MI=NULL. Based on the mutual information, EpiTag performs a greedy algorithm to iteratively select the subset of most informative SNPs to optimize power to detect epistasis.

A list containing the following components:

`tagSNPs.R1`	A `vector` that stores tagSNPs selected in `Region1`.
`tagSNPs.R2`	A `vector` that stores tagSNPs selected in `Region2`. Returned only if `Region2` is provided or if `mat.MI` contains the Mutual Information between two regions.
`pairs.tag`	A list of pairs of SNPs selected as a tag pair.
`pairs.bins`	A list of bins of SNPs pairs. Each bin is ix2 matrix where each row is a SNP pair tagged by the corresponding tag pair in the object `pairs.tag` and i is the number of such SNP pairs.

Friguet, C. and Emily, M. Selection de marqueurs pour la detection d'interactions de genes. Proceedings of 45emes Journees de Statistique, May 2013, Toulouse, France. pp.178, 2013.

getMatMI

############
## Example with simulated allelic data
############
data(RegionA)
data(RegionB)

## Computation of the mutual information and of the tag SNP selection simultaneously
res.EpiTag1 <- EpiTag(Region1=RegionA,Region2=RegionB,threshold=0.81,allelic=TRUE)

## Same results with a two steps computation of the mutual information and of the tag SNP selection
MatMI <- getMatMI(Region1=RegionA,Region2=RegionB,allelic=TRUE)
res.EpiTag2 <- EpiTag(mat.MI=MatMI,threshold=0.81,allelic=TRUE)


############
## Example with a real subset of genotypic data
############
## Loading of the genotypic data
load(system.file("extdata/Chr15_20S.Rdata", package="EpiTag"))

## Computation of the mutual information and of the tag SNP selection simultaneously
#### Can be long
res.EpiTag.real1 <- EpiTag(Region1= Chr15.20S,threshold=0.6,allelic=FALSE)


## Same results with the  precomputation of the mutual information
load(system.file("extdata/MatMIChr15.Rdata", package="EpiTag")) # Loading of the precomputed array of Mutual Information
res.EpiTag.real2 <- EpiTag(mat.MI=MatMI15,threshold=0.6,allelic=FALSE)