Skip to main content

Table 1 Type I error rates under different settings when the gene-gene correlation ranges from 0 (independence) to mild correlation (ρ=0.3), and when data were generated from GSE48091 to preserve the correlation in real application

From: Probabilistic prioritization of candidate pathway association with pathway score

  50 genes per set 100 genes per set
ρ=0 ρ=0.1 ρ=0.3 GSE48091 ρ=0 ρ=0.1 ρ=0.3 GSE48091
Bayesian 0.023 0.022 0.022 0.035 0.024 0.022 0.022 0.025
Logistic (ps) 0.043 0.045 0.046 0.066 0.041 0.040 0.041 0.060
Logistic (sum) 0.040 0.040 0.040 0.160 0.036 0.036 0036 0.041
GSEA 0.047 0.047 0.048 0.069 0.039 0.046 0.045 0.042
Global 0.031 0.030 0.033 0.131 0.047 0.046 0.050 0.159
ORA 0.096 0.094 0.091 0.069 0.057 0.058 0.060 0.085
Fisher’s 0.048 0.048 0.052 0.163 0.049 0.049 0.049 0.201
  1. The size of each set is either 50 or 100. The p-values under Global and Fisher’s are derived based on 1000 permutations