Fig. 1.
Indirect edge and RLowPC network. a An indirect association from X 1 → X 2 could arise from a regulatory structure of X 1 → X 3 → X 2. b RLowPC network inference. In an RLowPC network, firstly only the top ranked edges are kept in a pre-inferred PC network and then for each pair of genes, only the immediate neighbours will be regressed for PC calculation. In this example only the top 6 of 10 edges with highest correlations are kept and PC between X 1 and X 2 is re-calculated by removing the effects from two immediate neighbouring nodes (X 3, X 5). The correlation values are represented by the thickness of the edges