From: Predicting weighted unobserved nodes in a regulatory network using answer set programming
Iggy | MajS | |
---|---|---|
Instance (input) | \(\bullet\) An interaction graph, whose edges are directed and labelled as activation or inhibition. \(\bullet\) A list of discrete observa- tions on some IG nodes. | \(\bullet\) An interaction graph, whose edges are directed and labelled as activation or inhibition. \(\bullet\) A list of discrete observa- tions on some IG nodes. |
Search space/ Guess | \(\bullet\) Depends on node, sign \(\in\) {“ − ”, “0”, “ \(+\) ”} (3\(^{|V|})\) \(\bullet\) 1-influence repair added by node | \(\bullet\) Depends on node, sign \(\in\) {“ − ”, “0”, “ \(+\) ”} , weight \(\in\) [0, 100] (3\(^{|V|} \times 101^{|V|})\) \(\bullet\) K-influences repair added by node |
Logical rules | \(\bullet\) Experimental observation signs are kept. \(\bullet\) A node signed as “0” must receive only one influence signed as “0” or at least one “\(+\)” and one “−” in- fluence. \(\bullet\) A signed node must be jus- tified by at least one signed influence. | \(\bullet\) Experimental observation signs are kept. \(\bullet\) A node is signed “0” ei- ther if it only receives 0- influences or the same pro- portion of signed “ −”, “ \(+\) ” influences. \(\bullet\) A node is signed following the majority sign from all its received influences. |
Optimisation | \(\bullet\) Minimise the number of added repairs | \(\bullet\) Minimise the number of added repairs |
Projection (predicted nodes) | \(\bullet\) Six levels of possible pre- diction: 1 - 2 notPlus (-, 0) 3 0 4 notMinus (0, \(+)\) 5 \(+\) 6 CHANGE (\(+\), -) | \(\bullet\) Majoritarian sign \(\bullet\) Statistical information on the weight distribution (average, standard devia- tion) |