Abstract: Membrane protein is a kind of especial protein in its structure, which is of vital importance for living cells and play a role in communication and in the transport of the cells with the outside world. Due to difficulties of experimental techniques, only a handful of membrane proteins have been characterized. Therefore, theoretical prediction methods are important for structure determination of this class of proteins. In this paper, a new method based on the combining of the continuous wavelet transform(CWT) with the hydrophobicity of amino acids is introduced to predict the number and locations of helices in membrane proteins. 1YST is chosen from the protein database as an example to describe the prediction of the number and locations of helices in membrane proteins by using this method. The performance and factors influencing the predicted results are studied in detail. To further investigate the feasibility of this method, we randomly choose 36 proteins from protein database asthe test objects, there are 232 transmembrane helices(HTM), 222 of them can be predicted accurately, the average predicted accuracy is 96.1%, which indicates that this method is an efficient tool to predict the HTM of proteins.

Key words: Continuous wavelet transform, Transmembrane helices rigion, Hydrophobicity, Prediction