Atlas of Cancer Signalling Network (ACSN)

Atlas of Cancer Signalling Network (ACSN) maps and analyzes signaling pathway activities from gene expression profiles (GEPs) to identify known and novel pathway components in tissue- and disease-specific contexts.

Key Features:

• Integration of Prior Knowledge: ACSN incorporates prior biological knowledge about signaling processes to constrain and enhance pathway activity decomposition from gene expression profiles (GEPs).
• Deep Neural Network: ACSN employs a deep neural network to process complex GEP data and infer multi-variable pathway activities.
• Decomposition of GEPs: ACSN deconstructs gene expression profiles (GEPs) into multi-variable pathway activity signatures.
• Identification of Unknown Components: ACSN predicts previously unannotated pathway components and identified elements in mTOR, ATF2, and HOTAIRM1 signaling pathways from public GEP datasets.
• Validation and Cross-validation: Predictions are validated and cross-validated against regulatory information from genetic perturbation studies, transcription factor binding data, and external public GEP datasets.

Scientific Applications:

• Molecular Mechanism Discovery: Identifying signaling pathway components to elucidate molecular mechanisms of disease.
• Gene Prioritization for Experimental Design: Prioritizing genes for targeted experimental design based on inferred pathway activities.
• Pathway Analysis Across Tissues and Diseases: Comparing tissue-specific and disease-specific pathway activities by decomposing GEPs into pathway-level signatures.

Methodology:

ACSN integrates prior biological knowledge with a deep neural network to process gene expression profiles (GEPs), decompose them into multi-variable pathway activities, and validate predictions using genetic perturbation data, transcription factor binding data, and external public GEP datasets.