When to supsect flips after merging two datasets by plink?

Question

[khn]

Hello,

I have merged the two datasets (bed bim fam files by plink). One includes 20 cases, the other includes approximately 2000 controls. The results show like as below.

For example,

rs1234  chr3 A1 T A2 A CountA1 15/5/0, CountA2 350/890/660 

rs4567 chr3 A1 C A2 T CountA1 15/5/0, CountA2 370/870/660

rs8910 chr3 A1 C A2 A CountA1 15/5/0, CountA2 380/900/620

These all 3 snps surpassed the GWAS significance.

For example, db SNP for rs1234 said T/T 0.125, A/A 0.385, A/T 0.490. Ancestral allele is A.

Sometimes we need to suspect flips if no cases has major allele.

So, my question is, when to suspect flips? Then how to check whether it is really flipped or not?

Thank you in advance!