High-accuracy haplotype imputation using unphased genotype data as the references

• The accuracy is as high as 99.43%.
• It can finish a whole-genome imputation within 2 min on a laptop computer.
• The availability issue of ethnicity-matched references is solved.

Enormously growing genomic datasets present a new challenge on missing data imputation, a notoriously resource-demanding task. Haplotype imputation requires ethnicity-matched references. However, to date, haplotype references are not available for the majority of populations in the world. We explored to use existing unphased genotype datasets as references; if it succeeds, it will cover almost all of the populations in the world. The results showed that our HiFi software successfully yields 99.43% accuracy with unphased genotype references. Our method provides a cost-effective solution to breakthrough the bottleneck of limited reference availability for haplotype imputation in the big data era.