GRN

GRN infers gene regulatory networks from perturbation-based experiments to elucidate regulatory interactions involving the oncogene MYC in the squamous carcinoma cell line A431.

Key Features:

• Perturbation-Based Inference: Uses perturbation experiments to capture dynamic regulatory responses for network inference.
• Multiple Inference Methods: Infers GRNs using several computational inference methods to enhance robustness.
• False Discovery Rate Control: Controls false discovery rate using the NestBoot framework to minimize erroneous edges.
• Predictiveness Assessment: Implements a novel approach to evaluate the predictiveness of inferred GRNs against validation datasets in the absence of a gold standard.
• Cross-Validation and Independent Validation: Validates predictive performance through cross-validation benchmarks and independent datasets with different perturbation designs, outperforming null models.
• Novel Interaction Prediction: Predicts numerous novel regulatory interactions, with some experimentally validated.
• Data-Driven Approach: Performs inference solely from empirical perturbation data without relying on literature priors.

Scientific Applications:

• Mechanistic insight into oncogenesis: Captures regulatory interactions central to oncogenic processes to provide mechanistic insight into cancer.
• Discovery of disease progression mechanisms: Identifies novel regulatory mechanisms that may underlie disease progression.
• Therapeutic target nomination: Reveals candidate regulatory drivers that can serve as potential therapeutic targets.
• Comparative analysis of network alterations: Enables comparison of regulatory network alterations across perturbation designs and independent datasets.

Methodology:

Applies a computational inference framework to perturbation data from gene sets associated with MYC in A431, uses multiple inference methods, controls FDR with NestBoot, and assesses predictiveness via cross-validation, comparison to null models, and independent validation datasets.