Huge differences in variant calling between ExAC and 1000G

[Opio]

Hi,

I just performed my first variant calling on a Whole Exome Sequencing on 2 patients.
I did the pipeline following the recommendations and good practice from GATK, and at the end, with a MAF of ~2-3% (I tried both) on 1000G and an inhouse vcf, I have for some gene a huge difference between 1000G that gives me a frequency of 0.004 for a gene (MUC6) and for the same ExAC in NFE gives me a 0.4. So that's huge.

How can this difference could be possible?
I looked for databases in internet and also in databases in txt to look for any errors, but it seems not very relevant.

It is something sometimes happening?
What is the credibility of that kind of measures? One database says "rare enough" and the other one says " common" !

Best,

[Bukowski]

https://www.biostars.org/p/109425/

[Opio]

Hi Bukowski, thanks, we talked a bit about that in the group and I learned recently that it is highly polymorphic.
Still though, how can this be possible to have 0.004 in a database and 0.4 on the other one? It's comparing 40% versus 0.4% ! That's more about that part that I have troubles.
How ExAC and 1000G are built to have such differences in their references?

best,