engleski

New computational algorithm for accurate prediction of protein folding types

We present new algorithm for the prediction of secondary protein structures. The method enables evaluation of protein folding types from the nucleotide and amino acid sequences. Statistical and neural network testing of the model validity was performed on different proteins from the SCOP database, that posses defined folding types. The comparison of the algorithm to the methods of protein prediction based on hydropathic indexes and molecular resonance indicated that the algorithm shows considerable improvement in accuracy and specificity of the secondary structure classification. This procedure of computational analysis is not time consuming and does not require the use of computers with large processing capacity. If the algorithm is used together with self-organising neural network it performes almost immediate recognition of the secondary protein structure, which can be helpful in detecting type of protein folding from the newly sequenced exon regions. Moreover, graphical representation of the algorithmic results enables visual pattern recognition of different structures.

protein folding; secondary structure; algorithm; prediction

nije evidentirano