Naslov
New expert system for the prediction of protein folding types

Autori
Konjevoda, Paško ; Štambuk, Nikola ; Gotovac, Nikola

Vrsta, podvrsta i kategorija rada
Sažeci sa skupova, sažetak, znanstveni

Izvornik
Booklet of Abstracts of ICTCP-IV / Capron, Nathalie - Marly-le-Roi : International Society for Theoretical Chemical Physics, 2002

Skup
Fourth Congress of the International Society for Theoretical Chemical Physics (ICTCP-IV)

Mjesto i datum
Marly-le-Roi, Francuska, 09.07.2002. - 16.07.2002

Vrsta sudjelovanja
Poster

Vrsta recenzije
Međunarodna recenzija

Ključne riječi
protein folding; associative learning; compression algorithm; symbolic descriptors

Sažetak
The prediction of a secondary and tertiary protein structure is considered as the single most important step in further development of protein chemistry and rational drug design. Current systems for the protein structure prediction are homology based, and usually exploit beneficial effect of combining two or more predictive systems. We present a new rule based system for ab initio prediction of a secondary protein structure. The rules were extracted using algorithms for associative learning (APRIORI algorithm) on non-homologous database of protein structures (Jpred). The rules are based on fact that distributions and combinations of amino acid groups differ in particular elements of the secondary protein structure. The specificity of this present system is a compression algorithm that transforms original amino acid sequences to a limited number of symbolic descriptors, which improves the speed of analysis. The system is open, so the user can change the knowledge base and add new rules in accordance to his needs.

Izvorni jezik
Engleski

Znanstvena područja
Temeljne medicinske znanosti

POVEZANOST RADA