摘要

Identification of protein active sites is important to understand the function of protein and it is also crucial to the computer assistant drug design. In this paper, a unique method is presented to identify protein active sites based on time-frequency analysis by continuous wavelet transform (CWT). The numeric protein sequences are transformed by Morlet complex wavelet and protein active sites can be identified according to the energy scalogram. The results from hemoglobin and calcium binding protein indicate that the energy maximum of homologous proteins always appears in the same frequency, which is the characteristic frequency. In the energy scalogram, the time domain corresponding to characteristic frequency is the active sites of protein. The concentrated domains of energy are the critical domains of protein, which usually are conservative for homologous protein. Energy center can be utilized to identify active sites of protein. The proposed method has a potential to explore the function of protein.