Problem SNP that genotypes differently using different primers? - (Sep/18/2006 )

I have a novel A>G SNP in a candidate gene in the overlapping region between 2 fragments A and B thats giving very strange results. The 2 frags were sequenced in both directions, forward direction failed because of a poly A stretch upstream of SNP, the reverse sequence results were as follows in 46 samples: Reverse Primer_A - 41/46 were AA and 5/46 were GG; Reverse Primer_B - 41/46 were AA and 5/46 were AG (the same 41/46 were AA using both sequencing primers and same 5/46 were different using the 2 sequencing primers). It was noted that Reverse Primer_B had a SNP in its sequence but this was not in LD with the AG SNP in these 46 samples, also the sequence results were very clean using this primer throughout the fragment. Subsequent to these results I designed an internal sequencing primer pair for Frag A, to cover the A>G SNP in the 5 problem individuals in both directions. The 5 samples were called as GG homozygotes.
So i have 3 sequence primers , Reverse Primer_A and the internal primer pair calling the SNP as GG in the five individuals and Reverse Primer_B calling them as AG heterozygotes!!
I have talked to people in our group, our sequencing centre and also ABI helpdesk and they are all baffled. ABI have had a look at all my sequence files and have indicated its highly unlikely the SNP in Reverse Primer_B is causing the problem as you would expect more wholesale changes across the sequence instead of at a single SNP. I have repeated all of the sequencing above but nothing changes, I have even altered volume of BigDye but again, no changes.
I have attached a short word doc. showing the sequence data for one of the 5 problem samples so i'd appreciate any advice.

It is strange. Your "heterozygote" looks real enough from the chromatogram. The odds seem against it, but you may be the victim of allelic dropout with the primers that indicate GG, where one allele fails to amplify. I agree with your suspicions about H-W: you wouldn't expect so many GGs considering the frequency of the A allele. What creature are you genotyping? If human, you wouldn't expect too many SNPs so close to one another that could cause two of your primers not to work (your Reverse Primer A, and one of your new internal primers). I work with creatures that genetically are very highly variable, with SNPs everywhere, so I have to be very careful where I put my primers to avoid the null-allele problem. If the other 4 heterozygotes look like the one in your attachment, I'd suspect allelic dropout, even though it would seem unlikely.

Thanks for the reply, It is human DNA that i'm using . All the 5 heterozygotes are the same as the attached example which at least demonstrates some consistency in my results. I think I will have to go with the majority of sequence results and call the 5 individuals as GG homozygotes, even though highly unlikely, but when you consider it is only 46 samples its a possibility.