How does Plink choose which SNP is Allele1 and Allele2?

Question

[devenvyas]

I am was going back and double checking some old work, namely to confirm some Plink files had been merged properly (as when I was new to Plink I was doing some unorthodox stuff to get things to merge since my datasets had different IDs for the same SNP in thousands of cases).

Anyways, to my point, I've noticed at difference stages of my project across several BIM files, for any given chromosome coordinate, sometimes there are certain cases where Plink made one allele allele1 and then later made the alternate allele1

For one line once read

1 rs10915428 0 4158955 G T

but then after adding more samples to the dataset it became

1 rs10915428 0 4158955 T G

This made me want to know, how does Plink determine, which alleles becomes 1 and which becomes 2?

Answer 1

[chrchang523]

Plink 1.x normally sets the major allele to A2, and the minor allele to A1. It isn't really possible to do better with .ped + .map filesets which don't specify reference alleles.

Plink 1.9's --a1-allele and --a2-allele commands give you more control over this. --keep-allele-order also comes in handy often, though it's really annoying to have to remember to use it.

Plink 2.0 keeps track of REF/ALT alleles and flags "provisional" REF alleles (e.g. major alleles from a .ped + .map fileset).

[devenvyas]

So basically, whenever I add or subtract enough sequences for certain SNPs, the major and minor alleles could switch depending on which one is p > .5?

[devenvyas]

Also, is there anyway that one could set it that a given homozygous individual's allele is always allele1