I am working right now with VEP and I got SNPs from one external paper that has only the following information: snp, chr, pos, gene, risk allele, non-risk allele, frequency for a risk allele.

From MAF I can identify the major alleles and the minor allele. Sometimes a risk allele is a major allele. Example:

rs28411352    1    38278579    MTF1-INPP5B    T    C    0.26

Meaning that T is a risk allele and a minor allele.

rs12140275    1    38633879    LOC339442    A    T    0.78

Meaning that A is a risk allele but T is a minor allele.

For VEP annotation I need a reference allele, risk allele and a strand information. I have identified following cases:

I have looked up a bp in a reference hg19 and checked it with our major/minor/non-risk/risk.

CASE 1

rs10774624
>chr12:111833788-111833788
g
major:A minor:G non_risk:A risk:G

Here it means the reference bp is g, what corresponds to our minor/risk allele. => sense strand, ref = G, alt = A. (although G was a risk allele and a minor allele!)

CASE 2

rs9603616
>chr13:40368069-40368069
c
major:C minor:T non_risk:T risk:C

ref == major => sense strand, ref = C, alt =T (even if C is a risk allele)

CASE 3

rs331463
>chr11:36501787-36501787
A
major:T minor:A non_risk:A risk:T

It might mean, that we have an antisense strand => antisense strand, ref=T, alt =A

OR

that we have a sense strand (as in CASE 1) => sense strand, ref =A, alt= T

according to the dbSNP we have a sense strand, ref= A, alt =T

CASE 4

I did not find this case in my data but:

in fasta : A

major: T, minor: C, non_risk: C, risk:T (or major: C, minor: T, non_risk: C, risk:T)

I will assume then that we have an antisense strand => antisense strand, ref=T, alt =C

What do you think about it? Are those assumptions are legal if the strand information is missing?