Refining network reconstruction based on functional reliability

Reliable functioning is crucial for the survival and development of the genetic regulatory networks in living cells and organisms. This functional reliability is an important feature of the networks and reflects the structural features that have been embedded in the regulatory networks by evolution. In this paper, we integrate this reliability into network reconstruction. We introduce the concept of dependency probability to measure the dependency of functional reliability on network edges. We also propose a method to estimate the dependency probability and select edges with high contributions to functional reliability. We use two real examples, the regulatory network of the cell cycle of the budding yeast and that of the fission yeast, to demonstrate that the proposed method improves network reconstruction. In addition, the dependency probability is robust in calculation and can be easily implemented in practice.