ARACNE

ARACNE reconstructs gene regulatory networks from microarray expression profiles using an information-theoretic approach to identify direct transcriptional interactions in mammalian cells.

Key Features:

• Scalability: Scales to complex mammalian cellular networks.
• Data Input: Uses microarray expression profiles as input for network reconstruction.
• Information-Theoretic Filtering: Uses mutual information to detect statistical dependencies and filter indirect interactions.
• Direct Interaction Inference: Aims to distinguish direct transcriptional interactions from indirect associations.
• Validation: Validated on synthetic and real-world datasets.
• Accuracy: Can reconstruct networks exactly under conditions where loop effects are minimal.
• Resilience: Maintains robustness in the presence of numerous loops and mutual information estimation errors.
• Comparative Performance: Outperforms Relevance Networks and Bayesian Networks on synthetic datasets.

Scientific Applications:

• Transcriptional network reconstruction: Reconstructed transcriptional regulatory networks in human B cells.
• Oncogene target identification: Identified validated transcriptional targets of the cMYC proto-oncogene.
• Pharmacological target discovery: Aids identification of molecular targets for pharmacological interventions through inferred regulatory interactions.

Methodology:

Computes mutual information from microarray expression profiles and applies information-theoretic filtering to remove indirect interactions, enabling reconstruction of direct regulatory links and scaling to mammalian network sizes.