If the gene is significant in a survival analysis it should definitely affect survival length. Correlations is maybe not the best way of measuring this though.
Erling Häggström Gunfridsson I have read that the cox regression coefficient could determine the magnitude of relation between a gene's expression and the survival time. What are your thoughts on this?
The survival analysis is based on longitudinal time data. the expression of the genes should be correlated to the gene expression. The only thing I wonder about is the misunderstanding between Kaplan-Meier analysis and univariate Cox regression analysis. Are they the same? In fact, the Cox Proportional-Hazards Model is originally a multivariate statistical modeling.