Discrete Logic Modelling Optimization to Contextualize Prior Knowledge Networks Using PRUNET.

Area of Science:

• Systems Biology
• Computational Biology
• Bioinformatics

Background:

• High-throughput technologies generate vast biological data, often represented as influence networks or prior knowledge networks (PKNs).
• Existing PKNs may not accurately reflect specific biological contexts (e.g., cell types, disease states), leading to misinterpretation of new signaling data.
• Contextualizing PKNs is crucial for developing reliable, predictive models in cell signaling research.

Purpose of the Study:

• To present PRUNET, a user-friendly software tool for contextualizing PKNs to specific experimental conditions.
• To enable the identification of inconsistencies between measured data and signaling topologies.
• To facilitate the training of PKNs using context-specific datasets.

Main Methods:

• PRUNET takes a PKN and expression profiles of two stable steady states as input.
• It employs an iterative pruning process using an evolutionary algorithm for optimization.
• Optimization involves matching predicted attractors in a Boolean logic model with a Booleanized representation of phenotypes.

Main Results:

• PRUNET was validated on four biological examples.
• The resulting contextualized networks were used to predict missing expression values.
• Simulations of well-characterized perturbations were performed using the refined networks.

Conclusions:

• PRUNET is a tool for automatic curation of PKNs to match particular experimental conditions.
• The algorithm's general applicability supports various biological processes, including cellular reprogramming and health-to-disease state transitions.
• PRUNET enhances the reliability and predictive power of cell signaling models.