Hello,
I followed the steps for the Michigan imputation server and everything seemed to work well. My data was in build 38 so I had to lift it over to build 37 so that I could impute to HRC. I wanted to validated that the process was correct. I first ran a gwas of bmi which was small (600 people); the top hits were pretty random. I remembered that I had data on APOE4 carriage through direct genotyping so I thought this would be the best check. I noticed that the directly genotyped data doesn't match the APOE4 carriage as derived from the imputed data. For example, I would expect ID2 to be an APOE4 carrier given that ID1 is and that they have the same genotype. ID 6 should be a carrier? I assume that APOe4 is defined by having a C allele at rs7412 and one at rs423968. I wanted to check first before assuming everything is wrong.
Columns 6 and 7 refer to rs7412 and columns 8 and 9, I believe refer to snp rs423958. I extracted the snps using this code; plink --bfile ${SNP_DIR}/new_allchromosomes.converted.deanonymised.R2_0.3.MAF_0.05 --extract apoe_alleles.txt --recode alleleACGT --out f
imputed data ID1 0 0 0 -9 C T T C ID2 0 0 0 -9 C T T C ID3 0 0 0 -9 T T T C ID4 0 0 0 -9 T T C C ID5 0 0 0 -9 T T C C ID6 0 0 0 -9 C T C C
data from direct genotyping of APOE4 IID APOEGR2 ID1 Carrier ID2 Non-carrier ID3 Non-carrier ID4 Carrier ID5 Non-carrier ID6 Non-carrier
I would greatly appreciate any advice!
