Skip to main content
. 2013 May 31;8(5):e64832. doi: 10.1371/journal.pone.0064832

Table 2. Data used to construct both the initial and gold-standard networks used in the evaluation of PANDA and the other network reconstruction algorithms.

Network Name Data used to construct network[reference] Number of TFs/Genes/Conditions
Initial Cooperativity Network (Inline graphic) Affinity Capture-MS [47], [48] TFs: 53 (43 with evidence)
Gold Standard Cooperativity Network (Inline graphic) Low-throughput evidence [47], [48] TFs: 33
Initial Regulatory Network (Inline graphic) Motif [45], [46] TFs: 53 Genes: 2555
Gold Standard Regulatory Network (Inline graphic) ChIP-chip [45], [49] TFs: 52 Genes: 1073
Initial Co-regulatory Network – Knock-Out Data (Inline graphic) Gene Expression [43], [44] Genes: 2555 Conditions: 106
Gold Standard Co-regulatory Network (Inline graphic) ChIP-chip [45], [49] Genes:1073 (1072 co-targeted)
Initial Co-regulatory Network – Cell Cycle Data (Inline graphic) Gene Expression [51][54] Genes: 2555 Conditions: 56
Initial Co-regulatory Network – Stress Response Data (Inline graphic) Gene Expression [55], [56] Genes: 2555 Conditions:173

References include both the publication and the website from which the normalized data was downloaded. The number of transcription factors and genes reported are those used to construct each network. The 53 transcription factors and 2555 genes mentioned in the initial protein-cooperativity network, initial regulatory network and initial co-regulatory networks are the same set of TFs and genes for all networks, and represent those for which we had both motif and expression data. For the initial cooperativity network, we allowed TFs for which we had no Affinity Capture-MS data to be initialized as self-cooperating (Pii = 1). The transcription factors and genes used to construct the three gold-standard networks are the subset of the aforementioned 53 TFs and 2555 genes for which we had ChIP information (52 TFs and 1073 genes), or “low-throughput” interaction information (33 TFs). We used this subset of TFs and genes when evaluating the quality of each network. “Low-throughput evidence” data represents interactions with evidence from “co-fractionation”, “co-localization”, “FRET” or “reconstituted complex.”