hkb-survcompaper

Home Contact Details Research Teaching Software Links Projects

A comparative study of survival models for breast cancer prognostication based on microarray data:

does a single gene beat them all?

Software and Data

Paper

A comparative study of survival models for breast cancer prognostication based on microarray data: does a single gene beat them all? Haibe-Kains B, Desmedt C, Sotiriou C and Bontempi G in Bioinformatics, volume xx, number xx, pages xx-xx, in press, 2008. [Abstract] [PDF]

Breast Cancer Datasets

VDX dataset: This dataset is available from the GEO database (GSE2034) or from a R workspace (wang2005.RData).
TBG dataset: This dataset is available from the GEO database (GSE7390) or from a R workspace (transbig2006affy.RData).
TAM dataset: This dataset is available from the GEO database (GSE6532/GSE9195) or from a R workspace (loi2007.RData).
UPP dataset: This dataset is available from the GEO database (GSE3494) or from a R workspace (miller2005.RData).

R Package

The survcomp package is available from here. The package has five dependencies: survival, ipred and survivalROC can be downloaded from CRAN; ROC can be downloaded from bioconductor; prodlim can downloaded from Gerd's website.

Sweave Code

The Sweave code to generate all the figures and tables of the article: (i) using VDX as training set is avallable from here; (ii) using TBG as training set is available from here. Note that before running the Sweave code, you must put all the microarray data (i.e. wang2005.RData, transbig2006affy.RData, loi2007.RData, miller205.RData) in the mdata subdirectory.