by Wojciech Jaśkowski, Jacek Blazewicz, Piotr Lukasiak, Maciej Milostan, Natalio Krasnogor
Abstract:
Analysis and prediction of three dimensional (3D) protein structure become one of the crucial task in science nowadays. Unfortunately, it is hard to identify one methodology which gives the best prediction of 3D protein structure for different protein sequences. Trying to solve this problem, the concept of metaserver has been introduced. In this paper, we propose a new metaserver method called 3D-Judge, which uses an artifficial neural network (ANN) to select the best model from among models produced by individual servers. This decision is made basing on the mutual similarities between models produced by the servers and the knowledge obtained during the training. ANN is trained on historical data, e.g.,models from CASP experiment. Here, we compare 3D-Judge with 3D-Jury that is a popular and effective metaserver method. The obtained results indicate that the 3D-Judge is competitive to 3D-Jury and, in some cases, outperformes it.
Reference:
3D-Judge — A Metaserver Approach to Protein Structure Prediction (Wojciech Jaśkowski, Jacek Blazewicz, Piotr Lukasiak, Maciej Milostan, Natalio Krasnogor), In Foundations of Computing and Decision Sciences, volume 32, 2007.
Bibtex Entry:
@Article{Jaskowski2007_3djudge,
Title = {{3D-Judge} --- A Metaserver Approach to Protein Structure Prediction},
Author = {Wojciech Jaśkowski and Jacek Blazewicz and Piotr Lukasiak and Maciej Milostan and Natalio Krasnogor},
Journal = {Foundations of Computing and Decision Sciences},
Year = {2007},
Month = {jan},
Number = {1},
Pages = {3--14},
Volume = {32},
Abstract = {Analysis and prediction of three dimensional (3D) protein structure become one of the crucial task in science nowadays. Unfortunately, it is hard to identify one methodology which gives the best prediction of 3D protein structure for different protein sequences. Trying to solve this problem, the concept of metaserver has been introduced. In this paper, we propose a new metaserver method called 3D-Judge, which uses an artifficial neural network (ANN) to select the best model from among models produced by individual servers. This decision is made basing on the mutual similarities between models produced by the servers and the knowledge obtained during the training. ANN is trained on historical data, e.g.,models from CASP experiment. Here, we compare 3D-Judge with 3D-Jury that is a popular and effective metaserver method. The obtained results indicate that the 3D-Judge is competitive to 3D-Jury and, in some cases, outperformes it.},
File = {Jaskowski073djudge.pdf:jaskowski073djudge.pdf:PDF},
Homepage-url = {http://www.cs.put.poznan.pl/wjaskowski},
Keywords = {protein structure prediction, neural networks, metaserver},
Url = {http://www.cs.put.poznan.pl/wjaskowski/pub/papers/jaskowski073djudge.pdf}
}