paperKB
coga / coga-kb
Processing
Help
Sign in

Chunk #22 — Materials and Methods — MalariaGEN cross-validation experiments

Source
Genotype imputation with thousands of genomes.
Embedded
yes

Text

We also wanted to compare against Beagle on reference panels of primarily African ancestry, so we generated another reference panel called “HM3.afr” that included the HapMap3 haplotypes from the ASW, LWK, MKK, and YRI panels (822 haplotypes). We then used IMPUTE2 and Beagle to impute genotypes in the 1216 Gambians based on two reference panels: GMB (200 haplotypes) and GMB+GHN+HM3.afr (1222 haplotypes). We masked the SNPs as described above, except that this time we masked every 13th SNP so that Beagle would treat the data as it would a standard GWAS analysis. (Beagle invokes a different model-fitting strategy when fewer than 7% of the genotypes are missing from the study dataset.)