Survival analysis

Any single CpG loci that can be predictive of survival? 

From clinical_patient_public_OV.txt extracted vital status ("living" or "deceased"), days to the last followup days to death. Used this information to create a censoring indicator (thanks to Justin for the helpful discussion). 

Kaplan-Meier survival curve for TCGA ovarian cancer:

Residual disease as a predictor of survival. Obviously, patients with no microscopic disease show better median survival than patients with any leftover tumor. Although Charles Drescher was very interested in the patients where residual tumor is bigger than 2 cm because it looks very differently ("it is sticky"), it seems that patients have similar survival regardless of the size (if there is any left):

I also looked at the age and correlation in survival. it is possible that I made a bad choice in term of age groups but I don't see any significant differences (looking back at it I should have chosen similar sized groups). The top panel shows age distribution of OC patients.