Skip to main content

Table 2 The effect of choice of gene level statistic and of a corresponding transformation on the detection rate.

From: A general modular framework for gene set enrichment analysis

  no transformation squared rank squared
  t mod.t corr t mod.t corr t mod.t corr
set 1 0.94 0.94 0.94 0.63 0.67 0.74 0.84 0.84 0.84
set 2 1.00 1.00 1.00 0.90 0.92 0.95 1.00 1.00 1.00
set 3 0.00 0.00 0.00 0.00 0.01 0.01 0.01 0.01 0.01
set 4 0.79 0.81 0.83 0.40 0.44 0.50 0.54 0.54 0.54
set 5 0.95 0.96 0.95 0.34 0.41 0.44 0.38 0.41 0.38
set 6 0.00 0.00 0.00 0.86 0.85 0.88 0.95 0.96 0.95
set 7 0.01 0.01 0.00 0.99 1.00 1.00 1.00 1.00 1.00
set 8 0.00 0.00 0.00 0.70 0.70 0.75 0.74 0.75 0.74
set 9 0.00 0.01 0.00 0.81 0.90 0.89 0.80 0.82 0.80
   binary local fdr   
   t mod.t corr t mod.t corr   
  set 1 0.62 0.61 0.55 0.49 0.48 0.46   
  set 2 0.81 0.87 0.73 0.77 0.78 0.71   
  set 3 0.01 0.00 0.01 0.00 0.00 0.00   
  set 4 0.38 0.44 0.34 0.32 0.36 0.32   
  set 5 0.27 0.33 0.20 0.23 0.26 0.20   
  set 6 0.83 0.85 0.76 0.71 0.66 0.69   
  set 7 1.00 1.00 0.99 0.99 0.99 0.95   
  set 8 0.63 0.66 0.58 0.61 0.60 0.55   
  set 9 0.71 0.73 0.65 0.74 0.75 0.66   
  1. The values indicated are the proportion of significantly enriched gene sets (p-values ≤ 0.05) for various combinations of test statistics and their transformations, with fixed global statistic (= mean) and the use of resampling for computing the significance.