Multi-Site Assessment of ProteoRed Plasma Reference Sample for Benchmarking LC-MS Platform Performance

One of the missions of the Spanish Proteomics Network (ProteoRed ISCIII) is to assist its proteomics core facilities in evaluating their capabilities to perform qualitative and quantitative proteomics analysis. In 2010, the ProteoRed's Sample Collection and Handling Group designed a moderately complex plasma standard reference sample primarily to be used for routine quality assurance monitoring of laboratory instrumentation, as well as inter-laboratory performance assessment, and development and validation of novel technologies. The ProteoRed Plasma Reference (PPR) sample is a subset of highly abundant well-characterized human plasma proteins with a number of isoforms, in addition to 4 spiked-in proteins, altogether distributed over 5 orders of magnitude in concentration. The PPR sample was recently stress tested in the latest ProteoRed Multicenter Experiment (PME6) that counted with the participation of 17 proteomics facilities using a wide range of LC-MS platforms. Although each laboratory was allowed to use its own favorite methodology, we requested the sample be analyzed in a single LC-MS run in experimental triplicate (3 different digestions).Evaluation of the results submitted by the study participants revealed moderate discrepancies at the peptide identification level, and poor overlap at the protein identification level. In an attempt to identify the source of such irreproducibility, raw data of 8 laboratories (24 LC-MS runs) were reanalyzed centrally using a standardized data analysis pipeline, which included protein inference using ProteinProphet software. We found that the majority of protein identification discrepancies across submitted reports of these 8 laboratories were due to inconsistencies on how data analysts and computational tools group and/or infer proteins. Immunoglobulin variable chain identifications were particularly conflicting throughout identification lists, even in the centralized analysis. Using a series of LC-MS performance metrics, we benchmarked the performance of 8 LC-MS instruments (Orbitraps) and identified system components that vary the most across laboratories.