摘要：Liquid chromatography-mass spectrometry (LC-MS) can separate the organic components of samples, which is widely used for qualitative and quantitative analysis of pesticide and veterinary drug residues, environmental pollutants and allelochemi- cals in soil, water and air science research. It also an effective tool in the determination of non-volatile compounds, polar compounds, thermo-unstable compounds and macromolecular weight compounds including proteins, polypeptides, polymers, etc. In this paper, a new matching algorithm was proposed to solve the problems of low accuracy and coverage of peptide in LC-MS replicates spectrum. At present, the key is to match the LC peaks and analyze the differences when peptide signals are detected. Generally, most algorithms are based on time warping functions. However, the difference of elution time between replicate spectrum is randomly generated. Besides time feature, the isotope feature is introduced in this paper for building a classification model under the hypothesis that the same peptide obeys same isotope distribution in repeated experiments. This isotope classification model was generated and tested by training and testing peptide signal which was detected by LC-MS/MS. At last, the accuracy of this new model was over 95%, while 90% coverage rate showed that the isotope classification model was efficient.
