question about survival analysis

Question

[tujuchuanli]

Hi,

I have a set of genes (~1000 genes) which share a common feature. I want to perform survival analysis to identify prognostic genes from this gene set. I choose genes with Cox-pvalue<=0.05 and regard these genes as prognostic genes. However, someone told me that I should use the adjusted pvalue to determine prognostic genes.

Is it reasonable to use raw Cox-pvalue to determine prognostic genes? Or I must do the multiple correction?

[tujuchuanli]

Thanks, Kevin. you help me a lot!!

Answer 1

[Kevin Blighe]

Usually the log rank p-value is chosen, and I have seen situations where this is both adjusted and not adjusted for FDR. I go over a similar thing here: Survival analysis with gene expression

For a more 'intimate' response on adjusting p-values for Cox models, I would go to https://stats.stackexchange.com/