ISSN:
0006-3525
Keywords:
Chemistry
;
Polymer and Materials Science
Source:
Wiley InterScience Backfile Collection 1832-2000
Topics:
Chemistry and Pharmacology
Notes:
This report continues to explore the use of a strategy known as the antlion method for predicting polypeptide and protein structure. The method involves deformation of a biopolymer's potential energy hypersurface in order to retain only a single minimum, near to the native structure. The vexing multiple minimum problem thus is relieved, and the deformed hypersurface constitutes a key element in three-dimensional structure predictions with atomic resolution. In this more demanding pilot study, we provide evidence that the antlion method is capable of dramatically simplifying the surface of polypeptides by successfully predicting the native form of the naturally occurring 26-residue polypeptide melittin. The systematic hypersurface modifications employed in our previous work have been used again for this case, but have been supplemented by the output of a suitable neural network. This neural network involves a new feature: the use of amino acid biophysical scales for improving the secondary structure prediction accuracy of simple perceptrons. © 1993 John Wiley & Sons, Inc.
Additional Material:
4 Ill.
Type of Medium:
Electronic Resource
URL:
http://dx.doi.org/10.1002/bip.360330211