Paper Info
Title
DiANNA: a web server for disulfide connectivity prediction.
Abstract
Correctly predicting the disulfide bond topology in a protein is of crucial importance for the understanding of protein function and can be of great help for tertiary prediction methods. The web server http://clavius.bc. edu/similar to clotelab/ DiANNA/ outputs the disulfide connectivity prediction given input of a protein sequence. The following procedure is performed. First, PSIPRED is run to predict the protein's secondary structure, then PSIBLAST is run against the non-redundant SwissProt to obtain a multiple alignment of the input sequence. The predicted secondary structure and the profile arising from this alignment are used in the training phase of our neural network. Next, cysteine oxidation state is predicted, then each pair of cysteines in the protein sequence is assigned a likelihood of forming a disulfide bond-this is performed by means of a novel architecture (diresidue neural network). Finally, Rothberg's implementation of Gabow's maximum weighted matching algorithm is applied to diresidue neural network scores in order to produce the final connectivity prediction. Our novel neural network-based approach achieves results that are comparable and in some cases better than the current state-of-the-art methods.
Year
DOI
Venue
2005
10.1093/nar/gki412
NUCLEIC ACIDS RESEARCH
Keywords
Field
DocType
disulfide bond,multiple alignment,neural network,internet,proteins,protein sequence,secondary structure
Protein structure prediction,Biology,Protein sequencing,Algorithm,Software,Bioinformatics,Multiple sequence alignment,Artificial neural network,Genetics,Protein secondary structure,Blossom algorithm,Web server
Journal
Volume
Issue
ISSN
33
SUPnan
0305-1048
Citations 
PageRank 
References 
11
0.81
5
Authors
2
Name
Order
Citations
PageRank
F Ferrè119019.97
Peter Clote240047.41