paperKB
coga / coga-kb
Processing
Help
Sign in

Chunk #24 — Imputing Haplotypes into Untyped Individuals

Source
Detection of sharing by descent, long-range phasing and haplotype imputation.
Embedded
yes

Text

For untyped probands born after 1900, the average legacy coefficient is only 0.054, much less than the 0.176 of typed probands. This is because the untyped individuals are on average younger and there is some clustering in people chip-typed. Still, with 375,032 probands in this category, approximately 20,000 paternal and maternal haplotypes each could be imputed. For the 101,599 untyped individuals born between 1850 and 1900, the average legacy coefficient is 0.168. This corresponds to another 17,000 paternal and maternal haplotypes each. Another approximately 2000 paternal and haplotypes each could be deduced for individuals born before 1850. Overall, this corresponds to about 76,000 haplotypes being potentially imputable. The emphasis is on haplotypes instead of individuals because often only one haplotype of a person could be determined (e.g. M1 in Fig. 5). Also, while the 76,000 is an average number that applies to all locations in the genome with good coverage by the SNPs, who and which haplotype could be imputed vary from locus to locus. Finally, even when a haplotype was passed on to a typed child or grandchild, there are still instances where the haplotype cannot be reliably imputed. We believe that the chance of this is around 10%.