Ursgal, Universal Python Module Combining Common Bottom-Up Proteomics Tools for Large-Scale Analysis

• Published in: Journal of proteome research Vol. 15; no. 3; pp. 788 - 794
• Main Authors: Kremer, Lukas P. M, Leufken, Johannes, Oyunchimeg, Purevdulam, Schulze, Stefan, Fufezan, Christian
• Format: Journal Article
• Language: English
• Published: United States American Chemical Society 04.03.2016
• Subjects: Algorithms; computer software; Databases, Protein; High-Throughput Screening Assays; Peptides - analysis; proteome; proteomics; Proteomics - methods; Search Engine; Software; peptide identification; search engine; statistical postprocessing engine; combine search engine results; high throughput; Python
• ISSN: 1535-3893; 1535-3907; 1535-3907
• DOI: 10.1021/acs.jproteome.5b00860

Proteomics data integration has become a broad field with a variety of programs offering innovative algorithms to analyze increasing amounts of data. Unfortunately, this software diversity leads to many problems as soon as the data is analyzed using more than one algorithm for the same task. Although it was shown that the combination of multiple peptide identification algorithms yields more robust results, − it is only recently that unified approaches are emerging; , however, workflows that, for example, aim to optimize search parameters or that employ cascaded style searches can only be made accessible if data analysis becomes not only unified but also and most importantly scriptable. Here we introduce Ursgal, a Python interface to many commonly used bottom-up proteomics tools and to additional auxiliary programs. Complex workflows can thus be composed using the Python scripting language using a few lines of code. Ursgal is easily extensible, and we have made several database search engines (X!Tandem, OMSSA, MS-GF+, Myrimatch, MS Amanda), statistical postprocessing algorithms (qvality, Percolator), and one algorithm that combines statistically postprocessed outputs from multiple search engines (“combined FDR”) accessible as an interface in Python. Furthermore, we have implemented a new algorithm (“combined PEP”) that combines multiple search engines employing elements of “combined FDR”, PeptideShaker, and Bayes’ theorem.