Question about why Illumina HumanExome SNP arrays appear to detect [A/T] and [C/G] SNPs by using two different probes.

Question

[curious]

I am looking at the manifests for illumina SNP arrays like this one for HumanExome-12-v1-1:

ftp://webdata:[e-mail]/Downloads/ProductFiles/HumanExome/v1-1/HumanExome-12-v1-1-B.csv

If I understand correctly:

Most SNPs have one corresponding probe at a single address that is called ‘AlleleA_ProbSeq’. For these probes, the SNP of interest appears to be the next nucleotide that would be added on the 3’ end of this probe by the polymerase during the extension step.

I make the assumption that all A and T nucleotides used in the extension step of the assay will fluoresce red and all G and C nucleotides will fluoresce green, which is hopefully right.

If this is the case, then it makes sense why you would be able to differentiate [A/G], [C/T] SNPs with only one probe by examining green vs red intensity at that single address.

I noticed that SNPs that are [A/T] and [C/G] have two probes at two different addresses ‘AlleleA_ProbSeq’ and ‘AlleleB_ProbSeq’ .These probes appear to have the SNP of interest build into the 3’ end of the probe itself. Is the idea here that only DNA corresponding to Allele A will all bind to ‘AlleleA_ProbSeq’ and only DNA corresponding to Allele B will bind to ‘AlleleB_ProbSeq’? Then you could just differentiate what alleles are present by comparing intensity at the two different addresses?

Any guidance is appreciated, thank you.

Answer 1

[Kevin Blighe]

Hey Brett,

I'm not sure, as I am more familiar with the Affymetrix designs. The A/T and C/G probe combinations appear to be 'ambiguous', from Illumina's perspective, when considering the determination of TOP / BOTTOM (strand) and A / B alleles. There is a white paper from Illumina about it, here: “TOP/BOT” Strand and “A/B” Allele

You are correct on the fluorescent intensities, though, and how red-to-green / green-to-red ratio is used to determine the genotype.

Kevin

[curious]

Thanks for responding Kevin. I am still stuck on this, but will update with an answer when I figure this out. -Brett

[Kevin Blighe]

It also relates to the following fact: A/T and C/G SNPs are less common in the genome and are more difficult to detect (I believe) due to the fact that they are complementary bases. On the Illumina Infinium designs, these require 2 beads / probes. The other SNP types are non-complementary bases and thus require just a single bead / probe. So, the HumanExome array to which yuou refer is a mixture of these single- and 2-bead designs.

Probably a good idea to draw it out on paper and visualise the DNA strands separating, and the target probes coming in to bind to whatever is complementary to them.