[Bioinformatics methods and their comparative analysis of mass spectrometry]

The protein spectrometry holds such characteristics of complex and large volumes of data that the general statistical methods can't satisfy the demand of disease prediction or classification. Several kinds of main methods of mass spectrometry data mining,such as decision tree analysis, partial least squares, artificial neural networks and support vector machines is overviewed in bioinformatics perspective. And examples of different methods used to diagnose disease are illustrated. These show an important role of mass spectrometry in identification and prediction of disease.