Evert den Drijver

133 Genome-wide analysis in E. coli unravels homoplasy associated with cefotaxime resistance also located in genes annotated as being part of the type II secretion system. A complete overview is presented in Tables 2 and S4. Recombination analysis To verify whether the level of homoplasy could be a result of recombination, we used the Genealogies Unbiased By recomBinations In Nucleotide Sequences (Gubbins) algorithm to predict recombination events in our isolate collection (Croucher et al. 2015). This analysis showed frequent recombination events in our 172 E. coli isolates (see Fig. S3). Results illustrate that recombination blocks cover the region of the gspL and the thiM gene, and their high homoplasy levels could, thus, be due to recurrent recombination rather than independent mutations. Nonetheless, position −42 in the ampC promoter is not located in a region affected by recombination, as shown in Fig. S3. Moreover, when inferring the phylogenetic tree corrected for recombination events as obtained from the Gubbins analysis, the −42 C>T mutation actually occurred in 18 independent branches rather than the 17 branches in the uncorrected tree. This supports our previous result that this mutation is homoplastic, and not the result of a recurrent recombination event. Pyseer analysis To add additional support to our findings, we used Pyseer as an alternative method to compare the 24 positions identified with the homoplasy-based association analysis (see Table 2). Pyseer identified 65501 unique significant mutations associated with CTX resistance with filter P value ≤0.05 and 1097 unique positions with filter and likelihood ratio test (lrt) P value ≤0.05. Of the 24 positions identified with the homoplasy-based association analyses, we identified 8 positions also reported by the Pyseer method (see Table 2). Furthermore, the Pyseer method identified 6 complex mutation variants and a total of 14 positions that have multiple mutation variants overlapping the same genomic positions as found by the homoplasy-based method. The most dominant position associated with CTX resistance is the −42 C>T promoter mutation as indicated by both methods. No further positions on the promoter or the attenuator were found significantly associated with CTX resistance. 7

RkJQdWJsaXNoZXIy MTk4NDMw