An iterative and automated computational pipeline for untargeted strain-level identification using MS/MS spectra from pathogenic samples

Mathias Kuhring; Joerg Doellinger; Andreas Nitsche; Thilo Muth; Bernhard Y. Renard

The central step in the analysis of MS/MS spectra is an efficient peptide search for comprehensive protein identification. Therefore, we rely on established and reliable open-source database search engines such as X!Tandem 47 in combination with the XTandem Parser 48 and MS-GF+ 49 . However, any command-line search engine including proprietary ones could be implemented via additional Snakemake rules. We apply a classic target-decoy approach for false discovery rate (FDR) control 50 . Decoy sequences are created upfront and independently of the search engine with fasta-decoy.pl 51 by reversing the target sequences (including contaminant proteins, for instance cRAP 52 ) and both target and decoy databases are concatenated as suggested by Jeong et al. 20 . PSMs are subjected to an FDR cutoff based on a per-match FDR calculated as Ndecoy/Ntarget, with Ndecoy being the number of decoys in between targets (Ntarget) in a list of matches sorted by e-value 20, 53 . To acknowledge established false positive hits of previous iterations, decoy sequences with an e-value above the FDR-cutoff are passed on and concatenated to the database of the next iteration.