Analysis of Protein-Protein Interactions- A Computational Approach

In the past years protein-protein interactions have gained a strong interest inthe fields of pharmacy medicine biology and bioinformatics. In this bookstatistical information on protein-protein interactions are computationallycollected and analyzed. Characteristic properties are determined and theirpredictability estimated. Therefore the results from a common dockingapproach are re-evaluated with the collected information to discriminatestructures with high yet biologically meaningless geometric complementarityat the interface region from the near native structures. The results show thatalthough there is a noticeable improvement in the predictability afterapplying statistical information the overall accuracy is still low. To find othermore specific properties transient and permanent complexes were comparedto each other. The lack of data led to an extensive search for more suitablestructural data and the development of an extensive database. This databasewas ultimately used to retrieve a large number of protein properties thatwere automatically analyzed for their separation precision. A high accuracywas obtained in separating transient and permanent interactions based onthe combination of only four properties. Combining this information withcommon docking approaches based on geometrical complementarity maylead to satisfying sensitivities.