|Keywords:||protein docking; HADDOCK; integrative modeling; co-evolution; clustering; protein interactions|
|Full text PDF:||http://dspace.library.uu.nl:8080/handle/1874/307246|
The living cell is a formidable entity kept intact and functioning by a network of interactions carried out by protein molecules. As such, understanding this network, the interactome, is key to understand the cell itself. To dissect the fundamental properties of protein interactions, researchers use advanced spectroscopy and spectrometry methods, such as x-ray crystallography and nuclear magnetic resonance, respectively, to deduce the three-dimensional placement of each atom of each molecule. This so-called atomic structure requires substantial effort and time to obtain experimentally, but is a prerequisite, for instance, to develop compounds that inhibit the interaction, e.g. some antibiotics. As an alternative to experiment, computational modeling methods such as docking offer high-throughput/low-cost predictions of atomic structures of interactions. Despite flawed, docking models are extremely valuable as tools to help generate or discard working hypotheses. In the last decade however, modeling methods started to integrate experimental data with their powerful algorithms. This class of methods, data-driven or integrative docking, quickly became a favorite among the (computational) structural biology community as it provided means to model molecular systems far beyond the reach of any isolated experimental or computational method with reasonable accuracy. Current developments in low-resolution experimental structure determination methods, such as cryo-electron microscopy and small-angle x-ray scattering, set the stage for integrative modeling protocols that can combine several sources of data, bridge different resolutions and different types of information, and offer an holistic view of the biological molecular machines.