两种软件分析唾液多肽组结果的比较

以10个唾液样本为例，唾液经氧化石墨烯-磷酸镧纳米复合材料（LaGM）方法分离和高分辨串联飞行时间质谱进行多肽鉴定后，使用Peaks studio 8.5（PS）和Protein Pilot Software 4.0（PP）两种软件分别进行搜库分析后对比鉴定结果。研究发现，两种软件鉴定出的阈值以上肽段数量存在差异，PS软件通常能够鉴定出更多阈值以上肽段数目；两种软件鉴定出的阈值以上相同肽段在PS中占30%~60%，在PP中占60%~90%，即某些肽段只能被PS或PP一种软件所鉴定。但是，两种搜库结果在降解蛋白质数目上无明显规律，数目因样本而异。值得注意的是，如果以PP和PS的阳性结果相互参照，发现在一种软件中阈值以下的肽段，在另一种软件中也可能是在阈值以上，而且此概率与肽段的打分呈正相关。研究还发现，两种软件对不同长度肽段的鉴定有一定的偏好性：PS鉴定结果中短肽段较多，而PP软件可以给出更多的长肽段。比较而言，在PP中，短肽段易出现假阴性，在PS中，长肽段易出现假阴性，而信号的相对强度与阈值无明显的相关性。本研究结果表明，只使用一种软件进行分析，结果不能准确地代表多肽组的全部情况，两种软件都能提供另一软件鉴定不到的信息，将两种软件结果综合分析，能够得到更为全面的结果。

In this study, 10 saliva samples were used as data sets. After the saliva sample was separated by graphene oxide-lanthanum phosphate nanocomposite (LaGM) method and identified by high-resolution tandem time-of-flight mass spectrometry, Peaks studio 8.5 (PS) and Protein Pilot Software 4.0 (PP) were used for searching analysis and the identification results were compared. It was found that the number of peptides identified by the two softwares were different. Under the experimental conditions, not only the PS software could usually identify more peptides, but also the results given by the two softwares did not completely overlap, and the same peptides identified by the two softwares occupied 30%-60% in PS and 60%-90% in PP, which meant that some peptides could be only identified by PS or PP software, but at the level of degraded protein, PS method had no obvious advantages. It was worth noting that if the positive results of PP and PS were cross-referenced, peptides below the threshold in one software may be above the threshold in another software, and this probability was positive related to the score of the peptide. Further research also found that the two softwares had a certain preference for identifying peptides of different lengths. The identified results of PS had more short peptides, while PP software could give more long peptides. In comparison, in PP software, short peptides were prone to false negative result, but long peptides were prone to false negative result in PS software. And the intensity of the signal had no obvious correlation with the threshold. This study showed that it was not advisable to use only one kind of software to analyze or only use the same reference result to represent the entire peptidome. Both softwares could identify the information each other that could not be identified by the other software. Cross-referencing the two analysis softwares each other could improve the accuracy of the results.