Summary of Integrated mESC Genomic Data.

*Top Ranked Edges | Top 0.01% of Edges.
Notes: A total of 77 high-throughput datasets were collected from various public sources to create a compendium of mESC-specific data that included 992 conditions (e.g. columns in a microarray matrix) and ∼2.2 million data points (Table S3). These data were standardized and integrated into ∼6 billion gene/protein pairs, and used as evidential data to generate a predictive mESC-specific network focused on mESC self-renewal and cell fate. Datasets were weighted based on the amount of shared mutual information contained in each as compared to all evidential datasets used by the Bayes net. A low mean redundancy indicates the dataset is highly unique. As observed in other similar Bayesian network data integration efforts (including integration of human data), genetic and physical interaction data were the most reliable, but also the least common [11]. We strove to assemble a diverse and comprehensive set of mESC data that would provide the most coverage and be highly informative. Protein-DNA Interaction data included chromatin immunoprecipitation (ChIP) followed by microarray hybridization (ChIP-Chip) and ChIP followed by high-throughput RNA sequencing (ChIP-Seq). Top ranked edges were the 639 edges with a rank order of 1 and an inferred edge weight ≥0.9999 (Figure 3A, Table S11); the top 0.01% of the network consists of the 22,664 edges with an inferred edge weight ≥0.9966 (Figure 3B, dataset contributions to top 0.01% edges available at StemSight.org/stemdata.html).