Stratify the analysis with clinical variables

TCP-Lab / transportome_profiler

Code to profile the expression of the transportome

GNU General Public License v3.0

0 stars 0 forks source link

Stratify the analysis with clinical variables #15

Closed MrHedmad closed 3 months ago

MrHedmad commented 1 year ago

Clinical variables could be used to stratify the analysis with different cohorts all following a certain input. metasplit can handle the splitting of the cohorts based on this data, although it's unclear on what clinical variables to split on. Initial candidates include:

Patient's age;
Patient survival (binned);
Tumor grading;

MrHedmad commented 9 months ago

Given the clinical data provided by the portal, we can stratify by tumor stage the following:

BRCA;
COAD;
LUAD;
BLCA;
SKCM;
HNSC;
LUSC

With the exception of HNSC and LUSC, they are all the tumors proposed by prof. Cassoni. HNSC and LUSC are there to compensate the fact that TCGA does not have a "Skin squamous cell carcinoma" category.

MrHedmad commented 3 months ago

The analysis has issues with the quality and error rate of the output. To be safe, we chose to not do any additional stratification to avoid increasing false positives too much.