Uszkoreit Julian, Maerkens Alexandra, Perez-Riverol Yasset, Meyer Helmut E, Marcus Katrin, Stephan Christian, Kohlbacher Oliver, Eisenacher Martin
Medizinisches Proteom-Center, Ruhr-Universität Bochum, 44801 Bochum, Germany.
J Proteome Res. 2015 Jul 2;14(7):2988-97. doi: 10.1021/acs.jproteome.5b00121. Epub 2015 Jun 10.
Protein inference connects the peptide spectrum matches (PSMs) obtained from database search engines back to proteins, which are typically at the heart of most proteomics studies. Different search engines yield different PSMs and thus different protein lists. Analysis of results from one or multiple search engines is often hampered by different data exchange formats and lack of convenient and intuitive user interfaces. We present PIA, a flexible software suite for combining PSMs from different search engine runs and turning these into consistent results. PIA can be integrated into proteomics data analysis workflows in several ways. A user-friendly graphical user interface can be run either locally or (e.g., for larger core facilities) from a central server. For automated data processing, stand-alone tools are available. PIA implements several established protein inference algorithms and can combine results from different search engines seamlessly. On several benchmark data sets, we show that PIA can identify a larger number of proteins at the same protein FDR when compared to that using inference based on a single search engine. PIA supports the majority of established search engines and data in the mzIdentML standard format. It is implemented in Java and freely available at https://github.com/mpc-bioinformatics/pia.
蛋白质推断将从数据库搜索引擎获得的肽段谱匹配（PSM）与蛋白质关联起来，而蛋白质通常是大多数蛋白质组学研究的核心。不同的搜索引擎会产生不同的PSM，进而产生不同的蛋白质列表。对来自一个或多个搜索引擎的结果进行分析，常常因不同的数据交换格式以及缺乏便捷直观的用户界面而受阻。我们展示了PIA，这是一个灵活的软件套件，用于合并来自不同搜索引擎运行的PSM，并将其转化为一致的结果。PIA可以通过多种方式集成到蛋白质组学数据分析工作流程中。一个用户友好的图形用户界面既可以在本地运行，也可以（例如，对于较大的核心设施）从中央服务器运行。对于自动化数据处理，有独立的工具可用。PIA实现了几种既定的蛋白质推断算法，并且可以无缝合并来自不同搜索引擎的结果。在几个基准数据集上，我们表明，与使用基于单个搜索引擎的推断相比，PIA在相同的蛋白质错误发现率（FDR）下能够识别出更多的蛋白质。PIA支持大多数既定的搜索引擎以及mzIdentML标准格式的数据。它用Java实现，可在https://github.com/mpc-bioinformatics/pia上免费获取。
