jump3

jump3 infers gene regulatory network (GRN) topologies from time series gene expression data using a hybrid model-free/model-based approach that combines a formal on/off gene expression model with non-parametric decision trees called "jump trees".

Key Features:

• Hybrid Methodology: Combines a formal on/off gene expression model with model-free elements to balance interpretability and scalability.
• Jump Trees (Non-Parametric Decision Trees): Uses non-parametric decision trees to reconstruct GRN topology without relying on strong parametric assumptions.
• Scalability: Capable of handling large datasets involving hundreds of genes for extensive network inference.
• Interpretability: Produces structured, interpretable models through decision-tree representations of regulatory relationships.
• Predictive Capability: Supports out-of-sample predictions via its structured modeling framework.

Scientific Applications:

• In silico and synthetic network benchmarking: Evaluated on simulated (in silico) environments and synthetic networks to validate performance.
• Real-world biological data analysis: Applied to real biological datasets, including identification of regulatory interactions activated by interferon gamma.
• Studying immune-response mechanisms: Used to uncover gene interactions involved in immune responses and other complex biological processes.

Methodology:

Employs a formal on/off (binary active/inactive) gene expression model and a decision tree-based, non-parametric "jump trees" approach that systematically analyzes time series expression data to infer regulatory relationships and construct network topologies.