Suite of R programs for reducing dimension of large data set, then 
running Gehan lasso to compute regularized coefficient estimates and 
"risk score"

Brent A. Johnson
24 September 2008

Assume we have n observations and p predictors

In cox_scores.R: [Requires 'survival' package]
	cox.scores() = p independent logrank test statistics
	top.coxscores() = returns D=30 strongest predictors

In gehan_lasso.R: [Requires 'quantreg' package]
	gehan.lasso() = rank-based lasso estimation via Gehan loss
	cv.gehan.lass0() = cross-validate Gehan lasso

In logrank.R:
	logrank() = logrank test statistic

In test_logrank.R:
	illustration of how to use programs

In test_gehan_lasso.R():
	illustration of how to use programs

In ExampleKaplanMeier.R():
	illustration of how to draw KM curve in R