Impact of measurement noise, experimental design, and estimation methods on Modular Response Analysis based network...

Area of Science:

• Systems Biology
• Computational Biology
• Network Science

Background:

• Modular Response Analysis (MRA) reconstructs signalling networks from steady-state perturbation data.
• Biological data often contains noise from experimental variability and multi-step measurements.
• Understanding noise propagation is crucial for reliable network reconstruction.

Purpose of the Study:

• To systematically investigate the impact of noise on network reconstruction using MRA.
• To evaluate how noise in concentration measurements affects inferred signalling network structures.
• To provide recommendations for optimizing MRA-based network reconstruction.

Main Methods:

• An in silico study of MAPK and p53 signalling pathways was designed with realistic noise.
• Statistical concepts and measures were employed to assess accuracy and precision of inferred interactions.
• Network structures were evaluated based on simulated noisy steady-state perturbation data.

Main Results:

• Larger perturbations enhance accuracy in network reconstruction, even with non-linear models.
• A single control measurement is sufficient for reconstructing signalling networks.
• Using the mean of replicate measurements in MRA is recommended over complex regression strategies.

Conclusions:

• Optimizing MRA requires careful consideration of perturbation size and data processing.
• Recommendations are provided to improve the robustness and accuracy of signalling network reconstruction from noisy data.
• This study offers practical guidance for researchers using MRA in systems biology.