Additional file 1 of GenAPI: a tool for gene absence-presence identification in fragmented bacterial genome sequences

Additional file 1: Table S1. Best non-self hit of all PAO1 reference genome BLAST experiment where 25% of the gene was blasted against the whole PAO1 genome. Only genes longer than 150 bp are reported, only the alignments with 98% identity and minimum 25% gene length were reported. rRNA genes were excluded from report as their sequences are identical. Paralog sequences were defined with BLASTP. Table S2. Assembly statistics of the simulated P. aeruginosa dataset reads. Table S3. Assembly statistics of the E. coli dataset reads. Table S4. Performance test of GenAPI, Roary, PBGA, PanDelos, panX, EDGAR and SaturnV using (a) a simulated S. typhi dataset with complete genome sequences used in Roary publication [4], (b) a simulated P. aeruginosa dataset with partly assembled gene instances (fragmented genome sequences) and (c) a dataset from the long-term experiment with E. coli [19]. TP – true positive gene deletion calls (true positive absent genes); FP – false positive gene deletion calls (false positive absent genes); FN – false negative gene deletion calls (false negative absent genes). TP, FN, and TN were counted only for genes that showed variation between genome sequences to avoid inflation of numbers from large amount of non-variable gene sequences. (d) As one of the absent genes was shorter than 150 bp, it was excluded from GenAPI analysis; when the requirement for the minimum gene length was reduced to 100 bp, all 181 absent genes were successfully identified. Table S5. List of known absent genes in simulated P. aeruginosa dataset [23]. 0 - gene is absent, 1 - gene is present. Table S6. List of known deleted genes and their prokka annotations in E. coli dataset [23]. Figure S1. Detailed scheme of the GenAPI workflow.