Fig. 2: Power calculations obtained using the sub-sampling approach for the detection of AMR genes. | Communications Biology

Fig. 2: Power calculations obtained using the sub-sampling approach for the detection of AMR genes.

From: PowerBacGWAS: a computational pipeline to perform power calculations for bacterial genome-wide association studies

Fig. 2

Results of running GWAS power calculations applying the sub-sampling approach for the detection of known AMR genotype-phenotype relationships (binary phenotype). These plots show the sample sizes required to detect AMR causal genes of different AF and effect sizes (for which full heritability is assumed). The y-axis shows the power, calculated as the proportion of GWAS replicates in which the causal AMR gene is above the Bonferroni-corrected genome-wide significance threshold. The black and dotted horizontal line marks 80% power. Sample sizes are represented in the x-axis. The colour of lines denotes different AF whereas point shapes and line types effect sizes in odds ratio units. The power calculation results presented here are those for the species-wide populations, see Supplementary Table 1 for sample sizes required in both species-wide and single-clade populations. Pan-genome GWAS was run to detect acquired AMR genes in E. faecium (a) and K. pneumoniae (b) populations. A burden test GWAS was applied to M. tuberculosis (c).

Back to article page