Estimating the parameters of a model for protein-protein interaction graphs

Area of Science:

• Graph theory
• Systems biology
• Bioinformatics

Background:

• Yeast two-hybrid (Y2H) assays are crucial for mapping protein-protein interactions (PPIs).
• Stochastic graph models offer a framework for analyzing complex biological networks like Y2H data.
• Understanding network motifs is key to deciphering biological functions.

Purpose of the Study:

• To develop accurate approximations for graph cycles in Y2H network models.
• To identify characteristic network motifs for reliable parameter estimation.
• To lay the foundation for Bayesian analysis of Y2H networks.

Main Methods:

• Stochastic graph distribution modeling.
• Analysis of three-cycles and unchorded four-cycles.
• Parameter estimation using graph statistics.
• Model testing against experimental Y2H data.

Main Results:

• Accurate approximations for expected counts of three-cycles and unchorded four-cycles were derived.
• Unchorded four-cycles were identified as characteristic motifs in the proposed model.
• The count of unchorded four-cycles proved to be a reliable statistic for parameter estimation.

Conclusions:

• The characterized stochastic graph model provides a robust framework for Y2H network analysis.
• Unchorded four-cycle counts are valuable for parameter estimation in Y2H networks.
• This work facilitates the use of Y2H network models in Bayesian analyses to identify potential data errors.