Content: B2: From Correlation to Causality: Reasoning Methods for Dynamic Protein Interaction Networks

B2: From Correlation to Causality: Reasoning Methods for Dynamic Protein Interaction Networks

Molecular biology has changed into a data-intensive science with exponential data growth. The large-scale availability of sequence, structure, and interaction data holds great promises to improve our understanding of living organisms and to cure disease. In particular, the dynamic interaction of molecules is paramount since all cells perform their functions as complex interaction of proteins, DNA, RNA and other molecules. Recently, there has been substantial progress in making large-scale, even genome-wide interaction data available, in analyzing their structure and applying them to uncover disease mechanisms. Nonetheless, there are several pressing, open problems:

  • Most available interaction data are static snapshots, while a living cell is a dynamic system varying with time.
  • Most available interaction data are error-prone and cover only a fraction of all interactions.

The project develops formal techniques and systems for reasoning over incomplete and error-prone protein interaction networks. Our approach is based on influence graphs and on hybrid reasoning techniques rooted in Answer Set Programming. Multi-Context Systems provide a formal framework for integrating knowledge sources. The developed techniques will be applied and evaluated on concrete biomedical examples demonstrating how to reason over metabolic and disease networks.