How to handle "no call" on SNP imputation results

Question

[Babbala]

Hi, i just imputed some SNPs using IMPUTE2, i pruned them and then kept all the haplotypes with probabilities higher than 90% (thresholding was done using Plink1.90)

but what i get is that some SNPs that do not pass the threshold are undefined with the symbol ./. (instead of the normal 0/0, 0/1 or 1/1)

I don't know if what i am doing is right and i need some advice on how to handle them because i would like to keep as much SNPs as possible and not removing samples.

I am new about imputation so any feedback is appreciated.

Thank you.

[Kevin Blighe]

Can you please show the code that you have used for pre-phasing, imputation, and then post-conversion of the imputed data to VCF? ./. means that, while a variant at this position was called in one or more other samples in the dataset, in the sample with ./. there was not enough information to make any call, usually no aligned reads.