The centrality lethality rule in signed protein interaction networks

Περίληψη

Essential are the genes/proteins which are indispensable for the organisms. A lot of research has focused on the identification of essential genes/proteins because they are considered part of the minimal gene set; they are possible drug targets for pathogens and more knowledge about them will contribute to the improvement of therapeutic strategies for human diseases. The experimental procedures for the detection of essential genes are expensive, laborious and in most cases unfeasible. Hence scientists have created tools for their prediction from other data using computational approaches. The most important results have come from centrality indices in protein interaction networks which formed the centrality - lethality rule. According to centrality - lethality rule the higher the interactions of a protein the more likely for it to be essential. Since its introduction, this rule has been expanded to other centralities and many novel methods have been developed that integrate a variety of data. Despite all these advancements, the protein - protein interaction network has largely remained the same. For a better representation of protein interactions, additional information should be taken into consideration like activation/inhibition, direction and molecular function. In this work, we used the first large scale signed protein interaction network, which was constructed using protein interaction and RNAi screen data for D.melanogaster, to predict essential protein using centrality indices. This revealed that when a protein has many activation interactions it is more likely to be essential.

Keywords: essential gene/protein, centrality lethality rule, signed protein networks, systems biology, protein complex