MUPRED is a protein secondary structure and solvent accessibility prediction server based on a novel framework developed by Rajkumar Bondugula and Dr. Dong Xu at Digital Biology Laboratory, Computer Science Department, University of Missouri-Columbia. We use fuzzy algorithms and neural networks for our prediction.  For secondary structure prediction, we achieved a consistent three state accuracy of 80.4%, for solvent acessibility prediction, we achieved a mean absolute error of 15.3%, when measured on large independent test sets.

The server is provided as a free service to the research community. The data used to train the prediction system and test the system is available for download from the "Downloads" page. An executable of the prediction system that enables researchers to predict the structures on their local computers will be available soon. Executables for both Windows and Linux platforms will be provided.

Please click "Web Server" icon in the navigation pane to access the server. Please report problems if any by sending an e-mail to mupred@digbio.missouri.edu.

Reference: Bondugula Rajkumar, Xu Dong. "MUPRED: A Tool for Bridging the Gap between Template Based Methods and Sequence Profile Based Methods for Protein Secondary Structure Prediction", Proteins 2007, Feb 15;66(3):664-70.

 

Design by Rajkumar Bondugula, Digital Biology Laboratory, Computer Science Department, University of Missouri-Columbia, USA

Scripts by Pavankumar Bondugula, Arturo Guillen & Ryan Sextro. Last updated  18 Mar 2007.