Thanks genereader.
That's the one.
GMAF is the global minor allele frequency.
Minor allele frequencies vary across populations, so dbsnp gives you an average.
I think another helpful statistic they could include would be the frequencies in different populations.
There is some effort in this direction on dbsnp, though with some SNPs they only include small numbers in different populaitons.
With some rare SNPs, say 1 in a thousand, I would be much less worried about them if a population somewhere in the world had
a MAF of 10%. Given the current state of genetic technology one would expect that a SNP with a MAF of 10% in any large population
would be known to be either benign or damaging.
The GMAF was the main interest I had when making these files. I later realized that dbsnp also provided the identity of the minor allele, though
this was in another of its many file formats. Knowing which SNPs in a file are rare seems to be very useful information. I have used this as a sorting filter for
my exome scan. I tried to upload a spreadsheet version of my results which would unlock much of the helpful informaiton in my files without the need
for programming skills though the uploader did not like the file format.
For the 23andme file I thought that the GMAFs would be helpful in finding DNA relatives. If anyone on the forum could help me think this through it would
be greatly appreciated. My thinking is that if there is a DNA relative with a 5 cM match then if this relative also had shared a rare heterozygous match
then it would be likely that this match were true. Typically a 5 cM is considered to likely not be a true match. However, let's say that I have a heterozygous
SNP on the shared segment with a GMAF of 0.0001 (23andme does have such rare SNPs.) If I were to ask the possible match what their genotype is for this
SNP and they also reported it as heterozygous, wouldn't this suggest that the 5 cM match were real? Anyone?
If this makes sense then it would allow me to find more of the true matches with low cMs.
Edited by mag1, 20 May 2016 - 12:02 AM.