I imputed a set of around ~10k individuals using the sanger imputation server (1000 Genomes Phase 3 with eagle and PBWT). The autosomal data looks perfect - passed QC fine. However, I am was QCing the X chromosome data and I noticed for pretty much all the SNPs in females there is negative genotype values in the gen file format (e.g. 1 0 -1). Has anyone come across this before? From what I can tell online this indicates very poorly imputed data and should be excluded. Is this the case?
If so - is there any way to improve the imputation quality for the X chromosome? After QC, (SNPs were screened for geno 0.1 and maf 0.01), there was approximately 24k SNPs in the genotype file put forward for imputation.
Many thanks for your help!