ECCB

Poster Details

Reference	P-A20
Surname	Straßer
Firstname	Wolfgang
Affiliation	Upper Austrian University of App
Country	Austria
Title	Insilico Proteomic System: Combining Public and Private Protein Data for Computational Validation of Experiment Results
Coauthors	Doris Siegl, Thomas Kern, Johann Bauer, Kamil Önder
Abstract	The Insilico Proteomic System (ISP) combines publicly available proteomics data with experimental laboratory data for analysis and insilico validation of protein-protein interactions. Public data is mirrored automatically, whereas experimental data can be entered by designated users into a secured database area (demo account available at http://biomis.fh-hagenberg.at/isp/PrimosLabDemo).
Long Abstract	Motivation: In the proteomics research area a vast amount of data was published in publicly available databases in the last decade. The Insilico Proteomic System (ISP) reuses this knowledge pool for analysis and validation of experimental laboratory data. The system is made up of three main parts: local data storage, a background system responsible for automated task execution, and web front-ends for data administration and representation. As a first application an insilico validation of protein-protein interactions was implemented. For this approach detailed protein information and verified protein-protein interactions are required to computationally approve observed interactions received from yeast two-hybrid experiments in the laboratory at the Department of Dermatology, Paracelsus Medical Private University in Salzburg, Austria. The intense use of public information requires a local copy of the data realized through the ISP database. The database is extended continuously since the significance of the validation methodâs results increases according to the dimension of the public data pool provided. Sensitive experimental data of our laboratory is stored and administrated in a separate part of the database to guarantee its privacy. Access to this data is restricted to authorized staff members, whereas the public part can be searched without restrictions by using a public search engine. Designated users have the right to administrate experimental data through a separate web front-end. They can add protein sequences to the restricted database part, define interactions found in their experiments and submit validation requests for those interactions. Since interaction validation is based on time-intensive calculations of sequence similarities, the ISP background system distributes calculations to a computer grid on our university campus. The results are then transferred to the requesting user via e-mail and also made available online within the restricted area. Furthermore, the background system automatically manages importing and updating local public data. It is freely configurable regarding execution time and functionality and will be extended for further applications within this research area. Results: The public part of the system integrates data from biological sequence databases like SwissProt, TrEMBL and GenBank, and protein-protein interactions imported from the molecular interaction database BIND. By the time of this writing (May 2005), 1.6 million proteins and about 70,000 protein-protein interactions are integrated in the ISP database. The local public data pool will be broadened by adding further interaction databases in order to enhance the relevance of validation results. Currently, the computer grid calculating the interaction validation consists of 140 PCs. During instruction hours the tasks run as background processes, whereas they fully occupy the computers the remaining time. In the future all available campus computers are planned to be integrated into the grid. Validations on experimentally found interaction data of bullous pemphigoid antigen 2 (BPAG2) produced promising results compared to negative control datasets. Availability: Public data integrated into the Insilico Proteomic System can be accessed through a web front-end at http://biomis.fh-hagenberg.at/isp/Primos. At the moment, access to the private data administration front-end is provided for our laboratory staff only, but a demo version is available at http://biomis.fh-hagenberg.at/isp/PrimosLabDemo. Contact: primos@fh-hagenberg.at
Keywords	insilico validation, protein-protein interaction, Insilico Proteomic System, parallel computing, proteomics, experimental data, data integration
PDF

Developed by SoftActiva