Through analysis of known protein structures, it is possible to gain insight into the rules that dictate how proteins fold. However, the number of experimentally determined protein structures is large and growing rapidly, which makes even the categorisation of protein structure difficult to perform. Computational tools can ameliorate this process, through automated categorisation and analysis.
A team from the University of Bristol, led by Prof Dek Woolfson, have recently published an article on AtlasCC. This computational resource automatically analyses the PDB to find an important protein substructure called the α-helical coiled coil and uses graph theory enumerate all the possible and observed structures this fold can adopt. These data are made accessible using a user-friendly, interactive web application that enables users to browse the structures. The application also identifies regions of coiled-coil structure space that has not been explored by nature, indicating possible opportunities for de novo design.
Below is a link to the press release for a paper that provides an example of how modelling can compliment experimental techniques in the quest for new or improved drug candidates. In this case the subjects are the nicotinic acetylcholine receptor (nAchR) subtypes associated with smoking addiction pathways. Selective binding of drugs to inhibit or partially antagonise the alpha4/beta2 subtypes is believed to increase the efficacy and decrease side-effects of smoking cessation therapies such as Varenicline, targeting nAchRs. The modelling contributions in this paper included producing homology models for extracellular domains of the receptor subtypes (see image below), docking the cytisine derivative compounds into each of these and performing molecular dynamics simulations using GROMACS. Careful examination of the compound behaviour and binding interactions within the different receptor subtypes allowed a rationalisation for the observed in vitro binding affinities.