Send to

Choose Destination
Nucleic Acids Res. 2005 Jul 1;33(Web Server issue):W230-2.

DiANNA: a web server for disulfide connectivity prediction.

Author information

Department of Biology, Boston College, Chestnut Hill, MA 02467, USA.


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 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.

[Indexed for MEDLINE]
Free PMC Article

Supplemental Content

Full text links

Icon for Silverchair Information Systems Icon for PubMed Central
Loading ...
Support Center