Skip to main content

Table 1 AIRnet results for DREAM3 competition

From: Inferring gene regulatory networks from asynchronous microarray data with AIRnet

Values for 10-gene networks.
  score AUPR p-value average AUROC p-value average
empty network 1.1816e+00 8.5675e-03 5.0578e-01
trajectories 1.6298e+00 2.1759e-03 2.5279e-01
heterozygous 2.2401e+00 3.6441e-04 9.0845e-02
null-mutant 2.8198e+00 4.1550e-05 5.5198e-02
Values for 50-gene networks.
  score AUPR p-value average AUROC p-value average
empty network 2.4438e+00 2.6065e-05 4.9687e-01
trajectories 2.6700e+00 6.1865e-06 7.3901e-01
heterozygous 2.6207e+00 7.7215e-06 7.4297e-01
null-mutant 1.4152e+01 5.2984e-26 9.3634e-04
Values for 100-gene networks.
  score AUPR p-value average AUROC p-value average
empty network 5.2312e+00 6.8572e-11 5.0297e-01
trajectories 3.8264e+00 4.7395e-08 4.6923e-01
heterozygous 5.2881e+00 6.3523e-11 4.1762e-01
null-mutant 3.7911e+01 1.0263e-71 1.4694e-05
  1. AIRnet results for 3 network sizes from the DREAM3 competition. The empty network row shows values for graphs with 0 edges and provide a baseline to interpret the scores for other networks. The empty network scores were generated by submitting an empty file with no predictions to the DREAM score evaluator. The other rows correspond to the type of data used to infer the networks. In the score column, larger values are better. In the other two columns, smaller values are better. Scores reported are produced using an 80% threshold parameter for AIRnet.