Beyond our immediate goal of fast and accurate phasing, we envision that the primary downstream application of Eagle will be genotype imputation (via pre-phasing with Eagle followed by imputation with other software) in the UK Biobank and future population cohorts of similar or larger size. We have demonstrated the utility of Eagle within current imputation pipelines and the promise of this approach for use in future data sets (e.g., imputation using N≈150,000 reference samples). However, realizing this potential will require additional work. First, as currently implemented, Eagle is optimized for phasing array data and will need to be modified to phase sequence data. In particular, the method will need to be modified to incorporate additional information available from paired-end reads36 and from rare variants—which can greatly aid IBD-calling—while accounting for increased error rates. Simulations with increased genotyping error suggest that the Eagle algorithm is in principle quite robust to error (Supplementary Table 15), but additional tuning will undoubtedly be necessary. Second, an imputation algorithm capable of rapidly and accurately imputing pre-phased target samples using very large imputation reference panels will
