Strange behaviour of the NCBI remap API ?

Question

[f.abbondanza94]

Hi everyone!

I'm rather new to GWAS studies and I've just received a cohort (HumanOmniExpress-12v1, hg18) from colleagues to clean for a future GWAS.

I want to move to b37/hg19 using Remap from NCBI. I've carried out standard QC steps (maf 0.01, cal rate < 98%, heterozygosity check, hwe 1e-6) (~ 100k SNPs lost in total), transformed the file into VCF and split it by chromosome

I've run the remap API with default settings but I lose a ridiculous amount markers, from 50641 to only 16587 after mapping for chr1. I've then used the liftOver web interface with default settings to move the entire dataset from b36 to b37 and I got the following message

Conversion failed on 239832 records.

Do you have any idea which is the problem?

Thanks! Filippo