ausblenden:
Schlagwörter:
-
Zusammenfassung:
Introduction: Cross-linking coupled with mass spectrometry (XL-MS) has been recognized as an effective source of information about protein structures and interactions. Many methods and tools have been developed and reported for XL-MS identification through the last decade. Every tool applies different heuristics to cope with the quadratic search space inherent in XL-MS data analysis and uses their own model to estimate the False Discovery Rate (FDR). As part of the release of OpenMS 2.4 we introduce version 1.0 of the tool OpenPepXL and compare it to other commonly used tools for identification of non-cleavable cross-linkers on a diverse set of XL-MS experiments.
Methods: OpenProXL is a protein-protein cross-linking identification tool implemented using the open-source and well-documented OpenMS library (Röst et al. 2016). Like xQuest (Rinner et al., 2008) it can make use of labeled linkers to denoise spectra by comparing the spectra containing the light and heavy linkers. It can be used efficiently without any heuristics to reduce the search space for small datasets and has an optional fast pre-scoring algorithm to make analysis of large datasets feasible. OpenPepXL is part of the OpenMS proteomics pipeline that includes tools for labeled and label-free quantification. It can be installed on Windows, OSX and Linux and is compatible with most computing clusters and cloud services for large scale data analysis.
Results: “Kojak”, “pLink 2”, “StavroX”, “XiSearch”, “xQuest” and our own tool “OpenPepXL” were compared on several datasets from different laboratories on the same computational setup. Some of these datasets were from XL-MS experiments with proteins or protein complexes with known structures and were used for structural validation of the results. We found that the overlap of identified cross-links among the compared tools with an FDR of 5% is less than 50% for some datasets, but it increases with more stringent FDR cutoffs. A consensus approach using multiple tools could be promising in extracting a set of very confident identifications. We also found that it can pay off to analyze datasets without using heuristics to reduce the search space, if the dataset is small enough to make such an analysis feasible with the available computing power. OpenPepXL makes an exhaustive analysis more feasible for larger datasets through its efficient implementation and by being compatible with computing clusters and cloud services. Without using heuristics to filter out spectra and cross-link candidates before a full scoring, OpenPepXL reports about 10% more cross-links than the other tools without losing specificity.
Conclusion: OpenPepXL is an efficient and versatile XL-MS search engine with broad applicability and better sensitivity than other available tools.
