Negative values in Cox regression analysis

Question

[Amir.Asgary]

Hi there. I am working on gene expression data of a pan-cancer project. I am new in survival analysis and have a question about coxph function in R package "survival". I have z-score of my gene expression levels which contains negative and positive values. the question is ,can I use this data with negative values to perform univariate cox regression analysis? what about multivariate cox regression analysis? Thanks.

Answer 1

[Kevin Blighe]

Yes, you can use a dataset that comprises both positive and negative Z scores. I am assured that you understand what each value on the Z-scale means? - please see my comment, here: Survival analysis with gene expression

For Cox regression, the fitted residuals for each gene should also follow a normal distribution, which should be virtually guaranteed with a [statistically] powered and balanced [by group] dataset on the Z-scale.

Kevin

[Amir.Asgary]

Yes. I have seen your comment very useful. So thanks to you, I am now sure about my path in analysis. Thanks so much