Proteinler arasındaki etkileşimleri tahmin eden PRISM algoritmasının otomatikleştirilmesi

Protein - protein interactions among numerous proteins are the building blocks of the biological networks. To understand the function of a cell, it is necessary to examine biological processes occuring in the cell. Many biological processes in the cell depend on these biological networks, thus researchers in the field of biology always give attention to these networks and interactions. There are many experimental methods available to predict protein-protein interactions. Also, many computational methods have been developed for this aim, especially in recent years. In this project, we present a web server for the automation of a protein-protein interaction prediction algorithm. PRISM (Protein Interactions by Structural Matching) algorithm predicts interactions among various proteins by using structural and evolutionary similarity to known template interfaces. Our software system constructs a relational database of pre-calculated protein-protein interaction predictions, and the template and target structures used to calculate these predictions. It involves three sections which are templates, targets and predictions. Contents of the database can be queried using the appropriate PRISM web page. Also, visualization of these template and target structures are generated using Jmol plug-in in our web server. Also, another important aim of our web server is to run the PRISM algorithm from scratch when the protein structures in hand are not present in the relational database. In the case of such a situation, the steps of the algorithm are executed in an orderly fashion and all manual operations between these steps are eliminated. Computation of PRISM algorithm can take a lot time and effort. Using a 4-tier web database application architecture and a message queue system, many computations can run concurrently and users do not have to do any manual operations during the computation aside from input their protein structures. Since the manual operations are eliminated, users do not have to wait one step to finish to continue to the next step of the algorithm.