RESUMEN
Objective To investigate the associated protein markers in urine of patients with chronic renal failure (CRF) based on surface enhanced laser desorption and ionization time of flight mass spectrometry (SELDI-TOF MS)technique.Methods Urine samples were taken from 150 CRF patients and 50 healthy people,and investigated by proteomic techniques with H4 gene chip.Results Compared to healthy group,141 different protein peaks were identified within the range of 1000 to 20 000 M/Z in the protein map of CRF group,whose differences were all significant (all P＜0.01).The decision tree model for CRF urine was constructed after bioinformation analysis to significantly differentiate between CRF and healthy group.The accuracy rate,sensitivity and specificity of the decision tree model were 96.0％,100.0％ and 94.7％ respectively.In CRF group and healthy group,the different protein peaks in urine were identified to probably be 17 proteins with reference to SwissProt database.Conclusions Candidate protein markers in urine are screened and prediction model of CRF urine is established.The markers are identified with the database which provides a more accurate prediction and solid evidence for early diagnosis of CRF.