SuPred: Yapay Sinir A ğ lari ve Sakli Markov Model kullanarak Protein İ kincil Yapi Tahmin Yontemi

In this paper we are presenting a new method called SuPred, which is designed by combining two well known methods used for protein secondary structure prediction. The prediction obtained using only ANN contain some discrepancy and therefore it is a big necessity to employ additional methods to get biologically meaningful results. For this reason, we are using a 7 state HMM in addition to a ANN to correct inconsistent results in a meaningful way. As the input for ANN, we have only considered biochemical features of amino acids. The overall success rate we got using a dataset of 453 proteins is 68% and the success rate of coil prediction reached to 85%