Skip to main content

Table 1 Summary of method performances from artificial data analysis

From: Robust extraction of functional signals from gene set analysis using a generalized threshold free scoring function

  parameters (if any) fixed size, up-regulation fixed size, up and down-regulation varying size, up-regulation Varying size, up and down-regulation included to fig. 1 and 2
  w1 w2 Score A Score B Score A Score B Score A Score B Score A Score B  
GSZ-score 0 0.1 -0.0002 -0.1898 0.0867 0.184 0.0111 0.0837 0.1025 0.4976  
GSZ-score 0.1 0.1 0.003 -0.0896 0.0883 0.26 0.0095 0.0988 0.1003 0.4669  
GSZ-score 0.2 0.1 0.0048 0.0119 0.0891 0.3061 0.0065 -0.0085 0.0965 0.2528  
GSZ-score 0.5 0.1 0.0074 0.1567 0.0898 0.3006 -0.0031 -0.355 0.0837 -0.0515  
GSZ-score 0 0.3 0.0054 0.0485 0.0901 0.33 0.018 0.5143 0.1073 0.7216  
GSZ-score 0.1 0.3 0.0067 0.1327 0.0904 0.3466 0.0156 0.4358 0.1043 0.5892  
GSZ-score 0.2 0.3 0.0076 0.1892 0.0904 0.3427 0.0128 0.2623 0.1007 0.3901  
GSZ-score 0.5 0.3 0.0088 0.2547 0.09 0.2874 0.0041 -0.1618 0.0895 0.0725  
GSZ-score 0 0.5 0.0054 0.0485 0.0901 0.33 0.018 0.5143 0.1073 0.7216 X
GSZ-score 0.1 0.5 0.0081 0.2245 0.09 0.309 0.018 0.5787 0.1049 0.5659 X
GSZ-score 0.2 0.5 0.0086 0.2609 0.0898 0.2778 0.0154 0.3876 0.1016 0.3828 X
GSZ-score 0.5 0.5 0.0094 0.2919 0.0892 0.2115 0.0077 -0.019 0.0917 0.1538 X
t-test 0   -0.0165 -0.2968 -0.2707 -0.7744 0.0056 -0.1683 -0.2454 -0.8415  
t-test 0.1   -0.0143 -0.2494 -0.2646 -0.7142 0.0074 -0.0679 -0.2385 -0.7783  
t-test 0.3   -0.0121 -0.2058 -0.2546 -0.6561 0.0087 0.005 -0.2275 -0.7206  
t-test 1   -0.0094 -0.1588 -0.2324 -0.6022 0.01 0.0675 -0.204 -0.6582 X
t-test 3   -0.0079 -0.1024 -0.2065 -0.5425 0.0106 0.1249 -0.1775 -0.5926 X
KS    -0.0001 -0.0268 0.0462 0.0027 -0.062 -0.5063 -0.0439 -0.3352 X
modKS    -0.0164 -0.1558 -0.0231 -0.3216 -0.0449 -0.5979 -0.076 -0.4214 X
iGA    0.0075 0.0911 0.0951 0.2485 -0.0058 -0.453 0.0827 -0.0325 X
  1. Table shows average performance of various methods and parameters with artificial datasets. Scores A and B are explained in the main text. The five best methods are highlighted with bold and underlined in each column. The next five methods are underlined. Five weakest scores are represented in italics. Scores obtained with fixed class size represent the performance when the results are normalized with class specific permutation, whereas the results with varying class size show the performance without any normalization.