Skip to main content

Table 2 Performance comparison in simulated data with continuous outcomes

From: Significance testing in ridge regression for genetic data

    Shrinkage Parameter
    Approximate test Permutation test
    0.1 1 10 100 0.1 1 10 100
Individuals SNPs          
500 20          
   TPR 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000
   FPR 0.045 0.045 0.061 0.133 0.015 0.015 0.017 0.095
  100          
   TPR 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000
   FPR 0.056 0.054 0.071 0.141 0.015 0.018 0.024 0.074
  1000          
   TPR 0.100 0.500 0.900 1.000 0.000 0.200 0.800 1.000
   FPR 0.038 0.045 0.049 0.080 0.007 0.006 0.010 0.029
  ALL          
   TPR 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000
   FPR 0.318 0.071 0.068 0.069 0.019 0.019 0.020 0.020
1000 20          
   TPR 0.900 1.000 1.000 1.000 0.900 1.000 1.000 1.000
   FPR 0.043 0.043 0.087 0.137 0.013 0.013 0.034 0.096
  100          
   TPR 0.900 1.000 1.000 1.000 0.900 0.900 1.000 1.000
   FPR 0.051 0.052 0.060 0.108 0.023 0.023 0.019 0.062
  1000          
   TPR 0.700 0.700 1.000 1.000 0.400 0.500 0.900 1.000
   FPR 0.060 0.058 0.055 0.076 0.007 0.008 0.010 0.020
  ALL          
   TPR 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000
   FPR 0.166 0.155 0.110 0.071 0.015 0.015 0.015 0.017
5000 20          
   TPR 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000
   FPR 0.048 0.048 0.048 0.113 0.006 0.006 0.006 0.053
  100          
   TPR 0.900 0.900 1.000 1.000 0.800 0.900 1.000 1.000
   FPR 0.055 0.052 0.062 0.100 0.003 0.001 0.007 0.055
  1000          
   TPR 0.700 0.700 1.000 1.000 0.700 0.700 0.900 1.000
   FPR 0.046 0.046 0.045 0.060 0.006 0.007 0.008 0.014
  ALL          
   TPR 0.400 0.500 0.900 1.000 0.300 0.900 0.900 1.000
   FPR 0.026 0.027 0.029 0.042 0.007 0.007 0.007 0.009
  1. Performance comparison between a permutation test and the approximate test. Data are simulated genotype data with continuous phenotypes. Reported are proportion of true positive and false positive results at significance threshold α = 0.05. TPR = True Positive Rate, FPR = False positive rate. Results for each simulation scenario are averaged over ten replicates.