Protein Secondary Structure Prediction Based on Statistical Dictionaries

This study first gives the definitions and construction methods of statistical dictionary and similar list, and then proposes two novel algorithms, hybrid windows prediction (HWP) and similar searching prediction (SSP), for protein secondary structure prediction. Our methods not only implement the incremental learning to utilize the increasing protein structure data, but also achieve high prediction accuracies of overall Q₃ and SOV by the blind test on the dataset composed of 2825 protein chains.