Abstract: MATH/CHEM/COMP 2002, Dubrovnik, June 24-29, 2002

PREDICTION OF PROTEIN FOLDING TYPES FROM THE PRIMARY NUCLEOTIDE SEQUENCES

Nikola stambuk¹, Pasko Konjevoda¹, Nikola Gotovac², and Biserka Pokric¹

¹Rudjer Boskovic Institute, POB 180, HR-10002 Zagreb, Croatia

²Department of Radiology, General Hospital Pozega, Osjecka bb, HR-34000 Pozega, Croatia

We present new method that enables simple prediction of a and b protein folding types from the nucleotide and amino acid sequences. SMO machine learning algotithm for the support vector machines was used for the classification. Relative frequency patterns of particular codons proved to be sufficient for the accurate SMO based discrimination of a and b protein folding types. The method is not dependent on the physicochemical parameter analysis of the nucleotide or amino acid sequences. Overall prediction results and procedure cross-validations indicated that this procedure enables simple, quick and accurate prediction of the protein folds on a PC. The mehod may be used for detecting the type of the protein folding from the newly sequenced exon regions.