Skip to main content
. 2020 Nov 26;11:583013. doi: 10.3389/fimmu.2020.583013

Table 5.

Comparison of killer-cell immunoglobulin-like receptor (KIR)*IMP highest probability and EM-reduced haplotype prediction frequencies.

hap reference frequency KIR*IMP KPI w/ EM KIR*IMP - KPI w/ EM - KIR*IMP -
# reference reference KPI w/ EM
1 55.20% 71.86% 59.70% 16.70% 4.50% 12.17%
2
3 10.90% 12.57% 9.60% 1.70% −1.29% 2.95%
11 1.40% 1.47% 0.60% 0.00% −0.82% 0.85%
4 12.80% 7.49% 15.30% −5.30% 2.55% -7.83%
5
9 2.10% 3.41% 2.90% 1.30% 0.78% 0.52%
7 2.60% 0.33% 3.60% −2.20% 1.00% −3.24%
6 6.90% 1.80% 5.90% −5.10% −1.08% −4.05%
10
8 2.10% 0.60% 0.50% −1.50% −1.65% 0.12%
17 0.30% 0.00% 0.00% −0.30% −0.23% −0.03%
14* 2.40% 0.40% 0.00% −2.00% 0.00% 0.00%
18* 0.30% 0.07% 0.00% −0.20% 0.00% 0.00%
12* 0.80% 0.00% 0.00% −0.80% 0.00% 0.00%
mean 97.00% 100.00% 98.10%      

The table shows the comparison of the predictions between both methods as well as with reference European frequencies from Jiang et al. 2012 (column 2), which is the source of the haplotype numbers (column 1). KIR*IMP’s haplotype frequencies for the 1496 GoNL haplotypes are in column 3; some haplotypes are combined, as the haplotype numbers distinguish KIR2DS4 alleles. Column 4 contains frequencies for EM-reduced KIR probe interpretation (KPI) haplotype predictions Column 5 compares KIR*IMP frequencies with the reference, as column 6 does for EM predictions. Finally, column 7 compares the frequencies of KIR*IMP and the EM-reduced predictions. Haplotypes with a predicted frequency of 0 in both KIR*IMP and KPI are not shown. Haplotypes 14, 18, and 12 are in KIR*IMP’s set of reference haplotypes, but not KPI’s.