lpc-package: This package implements the Lassoed Principal Components...

In lpc: Lassoed Principal Components for Testing Significance of Features

Description

LPC scores are computed for each gene; the method borrows strength across genes and can result in more accurate gene scores than simpler statistics. In this package, the LPC method is applied by regressing Cox scores (survival outcome), two-sample t-statistics (two-class outcome), or standardized regression coefficients (quantitative outcome) onto gene expression eigenarrays, with an L1 constraint.

Details

Package:	lpc
Type:	Package
Version:	1.0.2
Date:	2013-12-15
License:	GPL (>= 2)

The main function is "LPC", which computes LPC scores for each gene. The matrix of gene expression data, a vector of outcome, and the outcome type must be passed in to this function.

Author(s)

Daniela M. Witten and Robert Tibshirani

Maintainer: Daniela M. Witten <[e-mail]>

References

Witten, DM and Tibshirani R (2008) Testing significance of features by lassoed principal components. Annals of Applied Statistics.

See Also

www.biostat.washington.edu/~dwitten

Examples

### Uncomment to run


#n <- 40 # 40 samples
#p <- 1000 # 1000 genes
#x <- matrix(rnorm(n*p), nrow=p) # make 40x1000 gene expression matrix
#y <-  rnorm(n) # quantitative outcome
## make first 50 genes differentially-expressed
#x[1:25,y<(-.5)] <- x[1:25,y<(-.5)]+ 1.5
#x[26:50,y<0] <- x[26:50,y<0] - 1.5
## compute LPC and T scores for each gene
#lpc.obj <- LPC(x,y, type="regression")
#lpc.obj
## Look at plot of Predictive Advantage
#pred.adv <- PredictiveAdvantage(x,y,type="regression",soft.thresh=lpc.obj$soft.thresh)
## Estimate FDRs for LPC and T scores
#fdr.lpc.out <- EstimateLPCFDR(x,y,type="regression",soft.thresh=lpc.obj$soft.thresh,nreps=50)
#fdr.lpc.out

Example output

lpc documentation built on May 2, 2019, 2:49 p.m.