Skip to main content

Table 1 Percentage of miss-called ancestry

From: Using multi-way admixture mapping to elucidate TB susceptibility in the South African Coloured population

  Percentage of total [IQR] Percentage of Ancestry [IQR]
Type of miss-call LAMP-LD RFMix LAMP-LD RFMix
San as Bantu 4.10 [1.65–9.76] 1.95 [0.88–3.55] 13.28 [5.12–35.16] 6.19 [3.13–10.45]
San as non-African 1.11 [0.05–9.58] 0.27 [0.11–0.66] 3.12 [0.18–36.09] 0.89 [0.36–2.05]
Bantu as San 0.08 [0.00–5.49] 0.04 [0.00–0.13] 0.32 [0.00–21.54] 0.14 [0.00–0.49]
Bantu as non-African 0.45 [0.02–8.63] 0.09 [0.02–0.23] 2.36 [0.07–33.44] 0.37 [0.09–0.92]
Non-African as San 0.14 [0.00–8.31] 0.09 [0.03–0.19] 0.42 [0.00–25.87] 0.25 [0.08–0.54]
Non-African as Bantu 0.95 [0.14–9.66] 0.18 [0.07–0.33] 3.09 [0.35–28.68] 0.47 [0.20–0.92]
  1. This table reports the interquartile range (IQR) of the percentage of SNPs that were miss-called by LAMP-LD and RFMix per each of the six possible miss-call categories. The known ancestry of a simulated data set of 1500 SAC chromosomes was compared to the ancestry called by the software program (chromosome 1). The median percentage of miss-called SNPs across all SNPs as well as the median percentage of miss-called SNPs across SNPs of that source ancestry are shown. San ancestry can for example be miss-called as either Bantu or non-African ancestry. The median percentage of all SNPs that were miss-called as such are shown in the second and third columns of the first two rows, and the median percentage of San SNPs that were miss-called as such are shown in the fourth and fifth columns of the first two rows. The mean proportion of San, Bantu and non-African ancestry in the simulated data set was 0.3342, 0.2772 and 0.3885 respectively. The difference in number of SNPs miss-called by RFMix, compared to the corresponding number of SNPs miss-called by LAMP-LD, were significant with p-values <2×10-16 for each of the six possible miss-call categories.