LD LASSO for Case Control Genetic Association Study

Share:

Description

Performs a modification of the fused LASSO for SNP subset selection in genetic association studies. The LD LASSO requires densely spaced SNP data and haplotype block structure.

Details

The ldlasso package is a collection of functions designed to apply the LD LASSO methodology as described in Younkin and Rao 2010. Functions usesful for displaying the results are also included.

Package: ldlasso
Version: 2.0
Date: February 1, 2011

Note

See the vignette 'ldlasso_vignette.pdf' in ldlasso package directory doc/.

For a package useful for creating haplotype block boundaries see MATILDE by Pattaro et al., available at http://astor.som.jhmi.edu/~gp/software/matilde/

The following function may be used to create block boudaries with MATILDE.

find.bounds <- function( block.obj, prob.threshold = 0.95, B = 1e4, frac = 2 ){ Xa <- as.double.snp.data(block.obj@gtdata)
X1 <- ifelse( Xa == 0, "A", "B" )
X2 <- ifelse( Xa == 2, "B", "A" )
X <- makeGenotypes(matrix(paste(X1,X2,sep="/"), byrow = F, ncol = dim(Xa)[2] ))
LD.obj <- LD(X)
null.X <- null.density(X)
nXr <- data.frame(null.X$x, null.X$yR)
names(nXr) <- c("x", "y")
matilde.obj <- matilde(LD.obj$r^2, nXr, B=B)
n <- length(matilde.obj$L)
b.vec <- apply(matilde.obj$b[(n/frac):n,], 2, mean)
block.cood <- as.numeric((b.vec >= prob.threshold))
return( list( matilde.obj = matilde.obj, prob.threshold = prob.threshold, block.cood = block.cood ) )
}

Author(s)

Samuel G. Younkin
<samuel.younkin@gmail.com>

References

Samuel Younkin, Joseph Nadeau, Robert Elston and J. Sunil Rao, "The Linkage Disequilibrium LASSO for SNP Selection in a Genetic Association Study of Late Onset Alzheimer Disease," Technical Report, 2010

See Also

ld_lasso_method, ld_lasso, plot_ldlasso, plot_beta, heatmap

Examples

1
2
3
4
5
6
7
  # Load example data
  data(ldlasso_example)
  ls()

  plot_ldlasso( ldlasso.obj )

  heatmap(ldlasso.obj)