Probability threshold Scenario    0.1  0.3  0.5  0.7   R = 2  No. of edges detected  285  208 202 199  T = 500  No. of true edges detected 198  198 198  198    Predictive value positive  69.5%  95.2%  98.0%  99.5%    No. of true edges with            | E (aij | aij ¹ 0,Y) - 0.4 | < 0.1)     198  198 198  198      | E (aij | aij ¹ 0,Y) - 0.4 | < 0.05)      182  182 182  182  R = 10  No. of edges detected  254  208 199 194  T = 15  No. of true edges detected 198  198 197  194    Predictive value positive  78.0%  95.2%  99.0%  100%    No. of true edges with               | E (aij | aij ¹ 0,Y) - 0.4 | < 0.1)        171  171 171  171        | E (aij | aij ¹ 0,Y) - 0.4 | < 0.05)       115  115 115  115   R = 6  No. of edges detected  52  16 8 2 T = 5   No. of true edges detected 38  14 7  2    Predictive value positive  73.1%  87.5%  87.5%  100%    No. of true edges with             | E (aij | aij ¹ 0,Y) - 0.4 | < 0.1)         38  14 7  2        | E (aij | aij ¹ 0,Y) - 0.4 | < 0.05)       22  10 7  0

Table 2: For the 100 variable network: the number of edges detected using different presence probability thresholds, number of these edges which are in the true network, the positive predictive value (their ratio) and number of edges in the true network whose posterior mean for aij is within a given tolerance of the correct value.