You have identified the problem with Sanger sequencing of a pcr product. If you have 2 polymorphic positions and 2 alleles A/t and C/G then you cannot distinguish between the possibilities A+C, A+G,T+c and T+g and it gets even more complex with increasing numbers of polymorphic positions. The answer is to clone the pcr product and sequence clones but Wang takes an alternative route and assumes that there is one very common haplotype and he subtracts this sequence from the mixed sequence to give a probable sequence for the other allele. This works when you are looking for linkage between a rare allele and a phenotype in your samples

Dear Paul Rutland . Thank you very much for your answer and time. Can you please help me see the methodology more carefully. I may have missed something and misinterpreted. Because, in the supplementary table (2013 Wang mmc2 ), the two alleles for the genes (here HLA A, B and C) are very variable haplotypes and not common. HLA-B*13:01 is a little common for B gene because it is allele of study. But this allele is common in cases only (not control) and A and C alleles have no "very common" haplotypes.

I am not the best person to answer this Divya Rana since I have little expertise in haplotyping but locally there may be common alleles due to inbreeding for geographical, religious, social or medical reasons that make one allele set common in the area local to the study while being variable worldwide or even nationwide.. In their paragraph "MHC typing" the authors refer to allele frequency databases ( one which has now moved and no longer accessible) and use the term"deducted" . This means subtracted ( not deduced which is different) which suggests to me that they have not cloned for sequencing but have sequenced with many polymorphisms and subtracted the common sequence for their social group(s) to obtain the other chromosome sequence. Unfortunately I am not qualified to comment on how valid this method is but it will be less accurate than cloning and sequencing

Paul Rutland Dear Sir. Your answer is still very helpful and logical. Thank you.

Another possibility Divya Rana might be to amplify and sequence the multiallele dna and get a mixed base sequence. Then with an ARMS pcr primer matched to just one base of the first polymorphism it may be possible to amplify just the one sequence containing the matching base at the 3' end of the primer and subtracting this from the mixed sequence would give the other chromosome sequence accepting that this would not work when the first polymorphic base is homozygous but it might be less work than cloning

Paul Rutland Dear Sir. I think I understood what you said. Thank you for your suggestions. But my study does not require me to find all of the alleles. I am only interested if my allele of study, which is HLA-B*1301, is homozygotic or hybrid (to find Hardy Weinberg equilibrium). For this I need to clone those human DNA samples which are positive, by allele-specific qpcr to distinguish the allele. Then I will just check if any one of the clones (may be 10 clones), generated through universal primers PCR, is negative for the allele (negative could mean hybrid or dizygotic).

One friend suggested RFLP.

Is there any simplified way?

Thank you for your time.

I am not familiar with the sequence of this allele Divya Rana but if there are not too many base changes then you could screen samples with allele specific pcr for the first change, discard all negative samples then use allele specific pcr again for the next change until you have a set which contains all of the required bases and then check by sequencing for hetero/homo zygosity