A1 allele and maf

[prabin gurung]

Hi Chris, 

A small confusion. According to what i have read from previous posts,  "Since February 2018, PLINK 2.0 --glm has been based on minor alleles." The output file normally has "REF", "ALT", and "A1" allele columns; the regression is based on A1. 

1) So can we say that for the above default setting,  A1 is the minor allele and the freq we get from cols=+a1freq will be the MAF? 

2) Also, if we force the regression to be based on alternate alleles instead by adding the 'omit-ref' modifier, then the freq given from cols=+a1freq might not always be MAF ?

Thank you. 

[prabin gurung]

3) Just to add another point, for the default setting as in 1), can A1 allele be REF allele ? I ask this because in my glm output (PLINK2), i have SNPs where REF=A1. 

[Christopher Chang]

1. Yes.

2. Correct, in that case the A1 allele frequency could be higher than 0.5.

3. Yes.

[prabin gurung]

Thanks Chris, 

Some more queries: 

4) So following up from point 3 above, if REF=A1, can i say that for this situation, the A2 allele will be the ALT allele (A2=ALT) for biallelics ? 

5) OR do you recommend cols=+ax ? 

6) If everything "yes" for point 4, can I say A1=REF as effect allele (EA) and A2=ALT as non-effect allele (NEA) ? 

Sorry, i am trying to do meta-analysis and the allele codings from different datasets  just confused me. 

TIA

[Christopher Chang]

4. That is correct.

5. It's up to you, either approach works.

6. Yes, A1 is always the allele that the regression is based on.