Skip to main content

Table 3 P-values comparing power by different methods.

From: Re-sampling strategy to improve the estimation of number of null hypotheses in FDR control under strong correlation structures

  ρ True π0 BY SampG.q SampS.q SampG.m SampS.m
Data-B 0.3 0.9 3.71E-17 1.10E-04 1.06E-04 2.06E-02 8.37E-03
   0.8 2.74E-31 3.93E-05 2.97E-05 1.94E-03 5.58E-04
   0.7 7.82E-34 7.72E-08 2.57E-08 3.20E-07 1.09E-07
   0.6 4.48E-48 8.31E-12 9.01E-12 2.78E-12 5.55E-13
  0.5 0.9 1.67E-08 4.95E-03 3.35E-03 3.73E-02 7.94E-03
   0.8 1.06E-13 7.70E-04 6.05E-06 4.06E-02 9.28E-04
   0.7 6.42E-15 2.14E-03 1.12E-04 1.60E-02 2.37E-03
   0.6 1.28E-17 5.55E-04 1.77E-06 2.99E-03 1.41E-04
  0.7 0.9 9.89E-03 8.65E-02 4.81E-02 6.76E-02 5.24E-02
   0.8 4.72E-06 2.38E-02 1.11E-02 2.89E-02 1.73E-02
   0.7 2.65E-08 1.53E-01 7.25E-03 2.85E-01 5.92E-02
   0.6 1.73E-06 1.19E-01 2.52E-02 1.25E-01 5.93E-02
Data-M   0.9 1.77E-53 9.78E-05 1.00E-06 3.66E-03 1.37E-05
   0.8 1.69E-62 3.05E-06 2.16E-10 3.52E-04 7.14E-08
   0.7 1.53E-74 5.27E-03 4.66E-08 3.86E-01 6.87E-03
   0.6 2.37E-80 2.05E-04 3.42E-10 2.49E-01 1.34E-05
  1. The actual power in gene selections across the 100 simulations under each configuration by these methods was compared to that of q-value method using one-tail paired t-statistics.