Fig. 3: Effect of degree of homoplasy on the power to detect SNPs obtained using the phenotype-simulation approach. | Communications Biology

Fig. 3: Effect of degree of homoplasy on the power to detect SNPs obtained using the phenotype-simulation approach.

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

Fig. 3

These plots show the sample sizes required to detect causal SNPs of different effect sizes (in odds ratio units, showed as different colours) and degrees of homoplasy (number of independent acquisitions, shown as different point shapes) when simulating binary phenotypes (full heritability assumed). The power calculation results presented here are those for SNPs of 10% MAF, in both species-wide (panels a, c, e) and single-clade populations (panels b, d, f), see Supplementary Table 4 for SNPs of different MAF. The power in Fig. 3e, i.e. for SNPs with 50–100 homoplasy steps in M. tuberculosis population, are particularly noisy due to the low number of SNPs in this population arising 50–100 times in the phylogeny (only 9 variants), which makes power estimates of such a small sample to fluctuate.

Back to article page