论文部分内容阅读
目的用表面增强激光解吸电离飞行时间质谱(SELDI-TOF-MS)和蛋白质芯片技术检测类风湿关节炎(RA)患者血清蛋白质指纹图谱,探讨基于人工神经网络的蛋白质指纹图谱模型对RA血清诊断标志物的筛选。方法用H4蛋白芯片结合SELDI-TOF-MS测定了141例血清标本的蛋白质指纹图谱,其中RA 90例,系统性红斑狼疮(SLE)20例,健康志愿者31名。将筛选出的血清蛋白质指纹图谱作为输入,建立人工神经网络预测模型,用随机抽取的93例标本(RA 60例,SLE 13例,健康志愿者20名)作为训练组,进行训练与交叉验证,并用另外测试组48例(RA 30例,SLE 7例,健康志愿者11名)的血清标本盲法验证该模型,同时与抗环瓜氨酸肽(抗CCP)抗体检测结果进行比较。结果利用从训练组得出的基于人工神经网络的血清蛋白质指纹图谱模型,对测试组的48例未知血清进行预测,结果显示,对RA检测的敏感性为90%(27/30),特异性为90.9%(9/11),阳性率为90.2%(37/41),明显高于抗CCP抗体检测结果。结论血清蛋白质指纹图谱可有效筛选RA血清中特异性蛋白标志物,基于人工神经网络的血清蛋白质质谱模型较以往传统方法具有更高的敏感性和特异性。
Objective To detect the serum protein fingerprints of patients with rheumatoid arthritis (RA) by using surface-enhanced laser desorption / ionization time-of-flight mass spectrometry (SELDI-TOF-MS) and protein microarray technology to investigate the relationship between RA fingerprinting Screening of things. Methods The protein fingerprints of 141 serum samples were determined by H4 protein chip combined with SELDI-TOF-MS, including 90 cases of RA, 20 cases of systemic lupus erythematosus (SLE) and 31 healthy volunteers. The selected serum protein fingerprints as input, the establishment of artificial neural network prediction model, with a random sample of 93 cases (RA 60 cases, SLE 13 cases, 20 healthy volunteers) as a training group, training and cross-validation, The model was further validated by another 48 cases of RA (30 RA, 7 SLE, 11 healthy volunteers), and the results were compared with those of anti-cyclic citrullinated peptide (anti-CCP) antibody. Results Based on the serum protein fingerprinting model based on the artificial neural network obtained from the training group, 48 unknown serums of the test group were predicted. The results showed that the sensitivity to RA was 90% (27/30), specificity Was 90.9% (9/11), the positive rate was 90.2% (37/41), which was significantly higher than that of anti-CCP antibody. Conclusion Serum protein fingerprinting can effectively screen specific protein markers in RA serum. The serum protein mass spectrometry model based on artificial neural network is more sensitive and specific than the traditional methods.