Protein-to-Protein Interactions: Technologies, Databases, and Algorithms

Studying proteins and their structures has an important role for understanding protein functionalities. Recently, due to important results obtained with proteomics, a great interest has been given to interactomics, that is, the study of protein-to-protein interactions, called PPI, or more generally, interactions among macromolecules, particularly within cells. Interactomics means studying, modeling, storing, and retrieving protein-to-protein interactions as well as algorithms for manipulating, simulating, and predicting interactions. PPI data can be obtained from biological experiments studying interactions. Modeling and storing PPIs can be realized by using graph theory and graph data management, thus graph databases can be queried for further experiments. PPI graphs can be used as input for data-mining algorithms, where raw data are binary interactions forming interaction graphs, and analysis algorithms retrieve biological interactions among proteins (i.e., PPI biological meanings). For instance, predicting the interactions between two or more proteins can be obtained by mining interaction networks stored in databases. In this article we survey modeling, storing, analyzing, and manipulating PPI data. After describing the main PPI models, mostly based on graphs, the article reviews PPI data representation and storage, as well as PPI databases. Algorithms and software tools for analyzing and managing PPI networks are discussed in depth. The article concludes by discussing the main challenges and research directions in PPI networks.