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

For the viral samples we used all viral NCBI RefSeq proteins (314156 entries, via Entrez on July 10, 2017) in the iterative workflow and additionally all viral non-RefSeq proteins (1987275 entries in total, via Entrez on July 11, 2017) for the unique-PSMs-and Pipasic-based strategies. Viral spectra were filtered beforehand using corresponding host proteomes (all isoforms) including the UniProt Homo sapiens reference proteome (93588 entries, UP000005640, May 23, 2017) for the cowpox sample and Gallus gallus Red jungle fowl reference proteome (29732 entries, UP000000539, May 16, 2017) for the bronchitis sample. For the iterative examination of the bacillus samples, all bacterial NCBI RefSeq proteins (65996005 entries, release 82) were downloaded via FTP (ftp.ncbi.nlm.nih.gov/refseq/release/bacteria/). Since downloading all NCBI Protein entries for bacteria (taxid 2) is impractical and requires too much time using the Entrez API, we used the NCBI Blast NR database as most comprehensive, common and readily available protein resource. This database can be obtained in fasta format via FTP (ftp.ncbi.nlm.nih.gov/blast/db/FASTA/, release 19.07.2017) and the bacteria subset (93760438 entries) was extracted with in-house scripts to create the databases for the unique-PSMs-and Pipasic-based strategies.