Here I will show results of the univariate analysis of correlation of DNA methylation with clinical traits such as survival, days to tumor recurrence, tumor stage, tumor grade, age, primary therapy outcome and tumor residual disease.
Most of the P value distributions look flat or just bad so I didn't see why I would calculate FDR for them, that's why that column is empty.
Survival
P value |
# of targets |
FDR |
|---|---|---|
0.001 |
5 |
|
0.005 |
46 |
|
0.01 |
107 |
|
0.05 |
859 |
|
Method: Cox Proportional Hazards model, P values = Wald's test. This distribution of P values concerns me. If I had no signal the distribution would be rather uniform rather than skewed towards 1. I think it might mean that I have some normalization issues (need to show this plot to Brig)
Days to tumor recurrence
P value |
# of targets |
FDR |
|---|---|---|
0.001 |
27 |
|
0.005 |
149 |
|
0.01 |
302 |
|
0.05 |
1543 |
|
Method: Cox Proportional Hazards model, P values = Wald's test.
Primary therapy outcome
Method: Kruskal Wallis test
P value |
# of targets |
FDR |
|---|---|---|
0.001 |
19 |
|
0.005 |
111 |
|
0.01 |
205 |
|
0.05 |
1128 |
|
Tumor grade
Method: Kruskal Wallis test
Why does the distribution look so weird?!!!
P value |
# of targets |
FDR |
|---|---|---|
0.001 |
2 |
|
0.005 |
25 |
|
0.01 |
86 |
|
0.05 |
798 |
|
Tumor stage
Method: Kruskal Wallis test
It looks weird!
P value |
# of targets |
FDR |
|---|---|---|
0.001 |
7 |
|
0.005 |
39 |
|
0.01 |
105 |
|
0.05 |
830 |
|
Tumor residual disease
Method: Kruskal Wallis test
P value |
# of targets |
FDR |
|---|---|---|
0.001 |
17 |
|
0.005 |
89 |
|
0.01 |
218 |
|
0.05 |
1129 |
|
Age at the time of diagnosis
Method: Spearman correlation
P value |
# of targets |
FDR |
|---|---|---|
0.001 |
1362 |
|
0.005 |
2361 |
|
0.01 |
3091 |
|
0.05 |
5933 |
|