Computational discovery of dynamic cell line specific Boolean networks from multiplex time-course data.

Area of Science:

• Systems Biology
• Computational Biology
• Bioinformatics

Background:

• Protein signaling networks are crucial for cellular regulation but are often static representations of dynamic processes.
• Understanding protein interactions and modifications (e.g., phosphorylation) is key to deciphering cell cycle progression and diseases like cancer.
• Existing large-scale networks lack predictive power and integration with experimental measurements.

Purpose of the Study:

• To develop and apply an efficient, parameter-free method for integrating time-series phosphoproteomic data into protein signaling networks.
• To model the dynamics of four breast cancer cell lines using inferred Boolean Networks (BNs).
• To highlight cell-line-specific commonalities and discrepancies in signaling pathways.

Main Methods:

• Combined logic programming and model-checking to infer Boolean Networks (BNs).
• Utilized multiple perturbation time-series data from four breast cancer cell lines.
• Applied the method to a large-scale protein signaling network from the HPN-DREAM Breast Cancer challenge.

Main Results:

• Achieved a Root Mean Square Error (RMSE) of 0.31 on test data, outperforming the challenge's best method (RMSE 0.47).
• Demonstrated comparable performance to top teams with an AUROC score of 0.77 when compared to the mTOR pathway.
• The method successfully inferred cell-line-specific BNs, revealing pathway commonalities and differences.

Conclusions:

• The developed methodology is an efficient approach for dynamic model discovery from time-course experimental data.
• The approach can identify potential experimental errors in large-scale signaling network data.
• The cell-line-specific BNs provide valuable insights into breast cancer signaling dynamics.