Protein fold-class prediction server for sequences of unknown structure

Martin Reczko reczko at web.de
Fri Mar 22 18:12:19 EST 2002


Synaptic is pleased to announce the avaliability
of the prediction server

Determination of expected protein folds - DEF Version 2.11

at

http://www.stepc.gr/~synaptic/def2.html

The DEF2-system (Determination of Expected Folds) performs
protein fold-class, approximate protein domain and
secondary-structure-composition predictions for
sequences of unknown structure.
The folds are classified into 48 fold classes.
The classes are defined in the 3D-ALI database of 
Pascarella and Argos, Prot. Eng. 5:121-137 (1992)

The percentage of correct classifications into 48 classes
for 125 novel test proteins is 86.4 % (the DEF1 system was ~80%
correct classifying into 42 classes). The prediction of the 4 classes
for secondary-structure-composition is correct in 90.4 % of the
test cases.

The DEF system is described in

Fold Class Prediction by Neural Networks
M. Reczko, H. Bohr, S. Subramaniam, S. Pamidighantam, and A.
Hatzigeorgiou.
In Protein Structure by Distance Analysis, Eds: H. Bohr and S. Brunak,
1994, IOS Press, Amsterdam, pp. 227-286.

The DEF data base of sequence based protein fold class predictions.
M. Reczko and H. Bohr.
Nucleic Acids Res. 1994 Sep;22(17):3616-9.

An update of the DEF database of protein fold class predictions.
M. Reczko, D. Karras and H. Bohr.
Nucleic Acids Res. 1997 Jan 1;25(1):235.

- Run DEF2 on your own system:

You can order the current version of DEF2 from Synaptic.
Send an email to mailto:synaptic_ltd at yahoo.de to get
information about the license fee.




More information about the Proteins mailing list