Fig. 3: Relationships between V. cholerae genetic changes and phenotype.
a, log-transformed odds ratios for the association of different genes and MGEs with rice-water stool and severe dehydration, adjusting for the presence or absence of other genes and MGEs, as well as the date and site of collection. The centre point indicates the calculated log-transformed odds ratio, and the error bars represent 95% confidence intervals; n = 1,617 biologically independent samples. b, Percentage of reads assigned to V. cholerae by Kraken in stool metagenomic data. Box plots indicate the median and interquartile range, with whiskers extending to the most extreme values within 1.5× interquartile range. Samples were categorized by the presence of ctxB7 and ctxB1, which, for this time period and location, delineated sBD1 and BD2; ctxB1+ samples were subcategorized on the basis of the detection of PLE1 and ICETET; n = 230 biologically independent samples. c, Frequency of ICP1 anti-defence systems detected in samples from Bangladesh, by year and by lineage of the co-sequenced V. cholerae. d, Proportion of samples in which the lysogenic phage K139 was present in the genome, classified by which genes in the ddmABC antiviral system were present. P values are indicated for the results of a multivariate logistic regression testing the association (Wald test; two-sided) between changes in the ddmABC locus and K139, adjusted for the covariates indicated in Fig. 3a (ICP1, pSA7G1, SXT-ICE, PLE, wbeT, VCA0455–VCA0459, hlyA, lineage, date and site). No adjustment for multiple comparisons was applied. e, Representation of PLE+ and wbeT+ V. cholerae among the total number of samples from Bangladesh; inferred export events and the number of samples outside Bangladesh descended from these export events, from 2003 to 2023.
