摘要

Nuclear receptors (NRs) are important transcriptional regulators in animals. They regulate different functions, such as lipid, reproduction, carbohydrate metabolism, fibrosis and metabolism. NRs form a category of phylogenetically evolutional proteins and have been separated into diverse subfamilies on account of their domain function. But for predicting the subfamilies, the preliminary step is to distinguish whether the protein sequence is a nuclear receptor or non-nuclear receptor. The sample with a pseudo amino acid (PseAA) composition representation of the protein sequence so as to incorporate a plentiful amount of protein sequence pattern information in order to increase the prediction precision for the classification. This article, which is based on the value of hydrophobicity, hydrophilicity, side-chain mass for sequence, we put forward a new percentage of method to predict types from protein sequences of subfamilies. Three percentages are on the base of the physical and chemical properties were collected from each of the protein sequences are made for their PseAA. It could testify by means of the jackknife cross-check method that the total successful rate are over 95%. The experimental results indicate that bioinformatics based on theory methodology can simplify and make experimental studies more intuitive.