Given the predominantly European composition of the HRC panel, the benchmarks on 32 CEU trio children were of primary interest, and we observed in these benchmarks that all methods achieved substantially improved accuracy using the HRC panel versus the 1000 Genomes panel (Figure 4 and Supplementary Table 5). For each choice of reference panel, Eagle2 achieved the lowest switch error rate, consistent with our previous results. For phasing small European cohorts, Eagle2 with the HRC panel provides a >2x improvement in accuracy over 1000 Genomes-based phasing: 1.36% (s.e. 0.04%) switch error rate versus 3.52% (0.06%) using SHAPEIT2 or 3.27% (0.06%) using Eagle2 with the 1000 Genomes reference (Figure 4 and Supplementary Table 5).
