Fig. 6.

Comparison of the called-based SFS (referred to as VCF: vcftools) and the directly estimated SFS (referred to as BAM: Score-limited). The SFS was constructed for 365 EUR individuals in the 1000 Genomes Project. (A) Shapes of the inferred SFS (shown in colors in legend). As the VCF file only contains sites that are inferred to be polymorphic, we only considered polymorphic sites for the SFS inferred from the BAM files and rescaled it so that all elements sum to 1. (B) Relative deviation of a fraction of sites with the derived allele count of 1–20. We computed the relative deviation of the SFS inferred from the BAM files compared with the SFS computed from the VCF file in each derived allele frequency bin $i/(2n)$. (C) Tajima’s D comparison