Skip to main content
. 2005 Jan 19;76(3):438–448. doi: 10.1086/428439

Figure 1.

Figure  1

Simulations for European (A) and West African (B) data sets for the 5q31 region, which demonstrate increased error in haplotype inference using an unstructured tagging approach compared with a contiguous-block–structure tagging approach. All data sets shown carry <20% missing data assigned at random to each SNP. Percentage error is shown for each SNP locus on the haplotype.