Human Genotyping Studies:
SNPs, Microhaplotypes, and Large Insertion/Deletions

Modern approaches to single nucleotide polymorphism (SNP) genotyping are on two scales. Large-scale research projects aim to build genome-wide assays of hundreds of thousands of SNPs in parallel, using very high-throughput microarray-based platforms. Smaller-scale projects focus on lower numbers of SNPs in thousands of subjects in parallel. In this article experiences are described with the Light- Typer Instrument.

The progression towards genome-wide SNP genotyping and the increasing availability of SNP data on public databases has resulted in the development of technology for very high-throughput SNP genotyping. However, even with the low cost per genotype call offered by some instrumental systems, they are still too expensive for use on large subject collections for population studies. For researchers interested in a particular candidate region, the goal may be to genotype SNPs at a high density within that region in several thousand individuals. For these several-SNP/many-subject approaches, a common genotyping methodology is the use of oligonucleotide binding assays [1, 2]. These rely on distinguishing between a probe which has bound to a perfectly matched sequence and one which has bound with a mismatch. We have previously demonstrated the potential of melting a single oligonucleotide from mismatch and perfect-match targets to determine genotypes at sites of single base variation using thermal ramp electrophoresis [3].

The LightTyper uses a similar approach, but relies on change in fluorescence in liquid phase rather than change in electrophoretic mobility as the probe melts.
Assay Design and Optimisation
The LightTyper Instrument uses either SimpleProbe probes or HybProbe probes (both from Roche Applied Science). In both cases, the allele-specific probe is designed to melt at around 60°C, and is positioned with the SNP in the central third of its length. HybProbe probe assays are based on the FRET (fluorescence resonance energy transfer) principle, requiring a second probe of higher melting temperature (Tm) that will remain bound after melting of the allele-specific probe and is positioned two base pairs (bp) away. This so-called anchor probe carries either an energy donor or a quencher, depending on the chemistry in use (we typically use fluorescein on the allele-specific probe and dabcyl on the fixed probe). PCR optimisations are as standard (MgCl2 and annealing temperature). LightTyper assay optimisation improves the ratio of target strand to nontarget strand achieved by asymmetric PCR, and increases the signal/ noise ratio in the reaction. The former is achieved by increasing the number of cycles (to 50 or more), and optimising the ratio of nontarget to target primer (usually 1:5 but it can be up to 1:20). As the cycle number increases, the availability of the nontarget primer decreases until most of the synthesis becomes single-strand. Signal/ noise ratio can be improved by careful titration of the allele-specific probe to minimise the number of free probe molecules at the end of PCR.
Software
The LightTyper software contains effective automatic genotype calling. However, when handling 384-well arrays the facility for manual checking and editing of calls is time consuming. We have developed a program using Microsoft Visual Basic for Applications (VBA) within Microsoft Excel (Figure 1). This imports the results from the LightTyper, including automatic calls, and groups them by genotype. The operator can then view and confirm all calls (typically at a rate of 2-3 calls/second). Altering a call simply involves typing the new code, while confirming a call requires a single keystroke.
Gene families can cause significant problems for SNP assay design. Figure 2 shows the results of genotyping an SNP in alcohol dehydrogenase 1C (ADH1C). This gene is highly homologous with ADH1B and ADH1A resulting in co-amplification of the nonpolymorphic homologous sequence. However, differences in sequence of the dabcyl (quenching probe) target result in premature melting of this probe from the homologs. As a result, there is a constant peak at 53°C, with the wild-type allele peak at 57°C appearing as a shoulder on the 53°C peak. These patterns were too complex for the automatic calling algorithm and had to be manually called. This problem can be avoided by selection of gene-specific PCR primers, as we have done in research on the growth hormone gene cluster [4].
Closely Spaced Markers - Microhaplotypes
"Microhaplotypes" comprising two SNPs within 10-15 bp can in some cases be assayed using the LightTyper Instrument. Figure 3 shows the patterns that can be obtained when two SNPs are located within the sequence but it can be up to 1:20). As the cycle number increases, the availability of the nontarget primer decreases until most of the synthesis becomes single-strand. Signal/ noise ratio can be improved by careful titration of the allele-specific probe to minimise the number of free probe molecules at the end of PCR.
Genotyping of a Large Insertion/Deletion Polymorphism
We have designed an assay capable of genotyping the large exon 3 insertion/deletion of the growth hormone receptor (GHR) gene [5]. While the LightTyper Instrument is designed primarily to genotype SNPs, this particular 2.7-kb deletion appears to be the result of homologous recombination between two 99% identical retroelements in the GHR gene. Our assay amplifies both insertion and deletion alleles with one pair of primers, resulting in two very similar PCR products, differing at one nucleotide. The large size of the insertion and the short PCR extension time minimises amplification of the potential long PCR product. This allows a standard melting-curve analysis (Figure 4a). For other large insertion/deletion polymorphisms, it may be possible to use a multiplexed LightTyper assay with a single forward primer (upstream of the deletion) and two reverse primers (one in the deleted sequence, one outside) and a pair of probes (also one in the deleted sequence, one outside). If the probes have different Tm, the assay should perform as a LightTyper SNP assay (Figure 4b).
We thank the UK MRC and Hope for funding, Sylvia Diaper for technical assistance, and Jason McKinney (Idaho Technologies) for expert discussion during beta testing.