Table 1.
Algorithm 1: List-SP.
Table 2.
Algorithm 2: CaSPIAN.
Table 3.
Algorithm 3: CaSPIAN given a known subnetwork.
Figure 1.
The gene regulatory network corresponding to the IRMA network, obtained using different algorithms.
Solid arrows denote true-positives and dashed arrows denote false-positives. True-negatives and false-negatives are not depicted in the figures in order to avoid cluttering; however, they can be easily obtained by comparing the true regulatory network and the inferred networks. Precision is denoted by , sensitivity by
, accuracy by
, and the
-measure by
.
Figure 2.
A gene regulatory network of HeLa cell genes, reconstructed using different causal inference algorithms.
Solid arrows denote true-positives and dashed arrows denote false-positives. True-negatives and false-negatives are not depicted in the figures to avoid cluttering; however, they can be easily obtained by comparing the known regulatory network and the inferred network. Precision is denoted by , sensitivity by
, accuracy by
and the
-measure by
.
Figure 3.
The SOS network reconstructed using Lasso, TLasso, Algorithm 1 (List-SP) and CaSPIAN.
Black solid arrows correspond to true positives and red dashed arrows correspond to false-positives. The numbers above edges describe their multiplicity ratios (MRs); in order to avoid cluttering, we did not plot the MRs for the results of Algorithm 1. Note that only links with MR at least are shown. Precision is denoted by
, sensitivity by
, accuracy by
and the
-measure by
.
Figure 4.
The network corresponding to IRMA obtained by applying Algorithm 3 to various subnetworks.
Parameters of this algorithm were chosen as and
and average gene expressions were used. Solid arrows denote true-positives and dashed arrows denote false-positives. Precision is denoted by
, sensitivity by
, accuracy by
, and the
-measure by
.
Figure 5.
The network corresponding to IRMA obtained by applying Algorithm 3 to various incorrect subnetworks.
Parameters of this algorithm were chosen as and
and average gene expressions were used. Solid arrows denote true-positives and dashed arrows denote false-positives. Green solid arrows correspond to the incorrect edges in the subnetwork. Precision is denoted by
, sensitivity is denoted by
, accuracy by
, and the
-measure by
.
Figure 6.
The HeLA network inferred by CaSPIAN with D = 4 and k = 7.
Solid arrows denote true-positives and dashed arrows denote false-positives. Precision is denoted by , sensitivity is denoted by
, accuracy by
, and the
-measure by
.