[BioC] output of coxfilter function in genefilter package

James W. MacDonald jmacdon at med.umich.edu
Thu May 24 17:00:02 CEST 2007

carol white wrote:
> Hi, I have fitted a univariate cox regression a data set and retained
> the genes whose p-value were less or equal to 0.05. the total number
> of retained genes is 2746. I also filtered genes based on the same
> condition (pval <= 0.05) on the same data set using coxfilter
> function in genefilter package and got  2648 genes, i.e 98 genes less
> than the number of genes that I got by univariate cox model. Should I
> have not got the same number of genes?

Not necessarily. It depends on what p-value you are using. The coxfilter 
function uses the likelihood ratio to compute a p-value. If you are 
using a Wald or Score test, the p-value may be different enough for some 
genes that they would pass your p-value filter, but not for the p-value 
returned from coxfilter.

> Moreover, I think the output of coxfilter is just a set of genes
> whose pval <= a threshold like 0.05 and it doesn't provide a gene set
> based on p-val ranking (ascending or descending) which I need.

You are correct - coxfilter is designed to give TRUE/FALSE indices to 
filter your genes.. However, it wouldn't take much to change coxfilter 
to simply return the p-values (or alternatively for you to emulate 
coxfilter with your own function that will do what you want).



> regards, carol
> ---------------------------------
> [[alternative HTML version deleted]]
> _______________________________________________ Bioconductor mailing
> list Bioconductor at stat.math.ethz.ch 
> https://stat.ethz.ch/mailman/listinfo/bioconductor Search the
> archives:
> http://news.gmane.org/gmane.science.biology.informatics.conductor

James W. MacDonald, M.S.
Affymetrix and cDNA Microarray Core
University of Michigan Cancer Center
1500 E. Medical Center Drive
7410 CCGC
Ann Arbor MI 48109

Electronic Mail is not secure, may not be read every day, and should not be used for urgent or sensitive issues.

More information about the Bioconductor mailing list