PIVS: Protein-protein interaction inference and visualization system using sequence-based homology search with DIP and BIND

Ki-Bong Kim1, Mi-Kyung Lee2, Seo Hwajung
1kbkim@bioinfo.smallsoft.co.kr, SmallSoft Co., Ltd.; 2mklee@bioinfo.smallsoft.co.kr, SmallSoft Co., Ltd.

We developed the system, PIVS, which can predict the protein-protein interaction (PPI) and visualize the PPI map on the basis of the homology search of input sequences with DIP (Database of Interacting Proteins) and BIND (The Biomolecular Interaction Network Database). DIP is the database which categorizes experimentally determined interactions between proteins and BIND is the database which contains not only interaction but also molecular complex information. The PIVS is an application with a client/server architecture, that runs on a LINUX server and is accessed via a JAVA-based application from any computer that has access to the server via Internet. The server consists of two main components. One corresponds to databases which include the public databases – DIP and BIND, and our own local database to be required for protein-protein interaction inference and related information retrieval. Our own local database contains various genomic and motif/domain-related information concerning entry sequences of DIP and BIND. The information has been collected from COG (Clusters of Orthologous Groups of proteins), GO (Gene Ontology), and InterPro (an integrated database of protein families, domains and functional sites) by means of the data mining as well as the search engines – BLAST and InterProScan. The public databases comprise 7 tables and our own local database 5 tables, which are closely linked by each other to provide systematic information concerning the input sequence to be analyzed by end-users. The other is BLAST-based sequence comparison module which maps the input sequences on the entry sequences of DIP and BIND to infer the protein-protein interactions. The client is a JAVA-based application with a graphic user interface. To visualize protein interaction map, VGJ (Visualizing Graphs with Java) package was employed, which is a tool for graph drawing and graph layout. In conclusion, PIVS is a very useful system for predicting the function and interaction of unknown protein sequence as well as for visualizing the protein-protein interaction map. In addition, it offers integral genomic and motif/domain-related information concerning unknown input protein sequence.