DREAM5 - Gene Network Inference

The DREAM5 transcriptional network inference challenge from Marbach et al. (2012) addressed the biological problem of reconstructing gene regulatory networks from high-throughput expression data. In it, over 30 computational methods were benchmarked on E. coli, S. aureus, S. cerevisiae, and simulated data to evaluate their ability to infer transcriptional interactions. All necessary challenge data, predictions, and consensus networks are provided in this record. Task: The dataset can be used to study causal discovery algorithms. Summary: Size of collection: Four gene expression datasets on Escherichia coli (4,511 Genes), Staphylococcus aureus (2,810 Genes), Saccharomyces cerevisiae (5,950 Genes) and in silico (1,643 Genes). Task: Causal Discovery Problem Data Type: Mixed Data Dataset Scope: Collection of Datasets Ground Truth: Unknown Graph / Known Graph Temporal Structure: Static Data License: CC BY-ND 4.0 Missing Values: No Missing Values Missingness Statement: There are no missing values. Files: Text_and_Figures.pdf: The supplementary notes to the Nature Methods article. It contains (A) the DREAM5 network inference challenge description, (B) the gene expression compendia, (C) the gold standards and (D) methodological insights and analyses. 1_Challenge_Data_Supplement.zip: This directory contains the input data, gold standards, and true gene names for the four networks of the challenge. In addition, the Matlab scripts used for the prediction assessment are included. 2_DREAM5_method_scores_Supplement.xls: This .xls file summarizes the AUPR, AUROC and overall score of the competitor methods. 3_DREAM5_network_predictions_Supplement.zip: The complete set of network predictions submitted by the 29 challenge participants. Predictions from the 6 off-the-shelf methods and the community integration are also included. These predictions were evaluated using the gold standards and scripts provided in Supplementary File 1. 4_EColi_SAureus_SCerevisiae_all_predictions_Supplement.zip: Community networks for E. coli, S. aureus, S. cerevisiae, and the in silico compendium, obtained by integrating the predictions of all 29 submissions of the DREAM5 network inference challenge. 5_EColi_SAureus_SCerevisiae_50_perc_precision_Supplement.zip: This folder contains the E. coli and S. aureus community networks at 50% precision cutoff as a .cys Cytoscape Seccion file. 6_EColi_SAureus_network_modules_Supplement.xls: The enriched list of network clusters from the E.coli network predicted to have greater than 50% precision. 7_EColi_experimental_support_Supplement.xls: E. coli experimental support for tested interactions.