huwenboshi / ImpG

Gaussian imputation of GWAS summary statistics
http://bogdan.bioinformatics.ucla.edu/software/impg/
14 stars 5 forks source link

alleles in Z score file #4

Open alessandrotestori opened 7 years ago

alessandrotestori commented 7 years ago

Hello! I have noticed that while running ImpG there might be problems when alleles in my input Z-score file (ZF) differ from the provided mapping file (MF). Am I right? Basically, I have four cases: 1 - Alt allele and Ref allele match in my ZF and in the MF; 2 - Alt allele and Ref allele are inverted in my ZF with respect to the MF (e.g.: Alt allele in MF = A; Ref allele in MF = G; Alt allele in ZF = G; Ref allele in ZF = A); 3 - Alt allele and Ref allele in my ZF are the complement of Alt allele and Ref allele in the MF (e.g.: Alt allele in MF = A; Ref allele in MF = G; Alt allele in ZF = T; Ref allele in ZF = C); 4 - Alt allele and Ref allele in my ZF are inverted and the complement of Alt allele and Ref allele in the MF (e.g.: Alt allele in MF = A; Ref allele in MF = G; Alt allele in ZF = C; Ref allele in ZF = T);

If I understand correctly, I can still use ImpG, provided that I make the alleles in my ZF consistent with the alleles in the MF. When flipping or inverting, however, I also should invert z-scores in some cases (case 2 and 4): am I right?

Please, let me know. Thanks, best regards

huwenboshi commented 7 years ago

Yes, you are correct. Also, you should also filter out strand-ambiguous SNPs, e.g. SNPs with reference and alternative alleles (A, T) or (C, G).