Although the most common sequence variants are heterozygous, it is also important to identify homozygous variants. In general, scanning methods do not identify homozygous changes unless the unknown samples are mixed with wild type before analysis. High-resolution melting analysis is an exception to this rule as many homozygous variants can be identified directly (Gundry et al., 2003, Wittwer et al., 2003, Dobrowolski et al., 2005). For example, the A>T base change of HbS can be detected in the homozygous state (sickle cell disease) even though the Tm shift is only 0.1-0.2°C. Sometimes the shape of the melting curve is sufficiently different that homozygous variants can be picked up on standard scanning plots (difference plots of normalized, temperature-shifted data). This occurs, for example, when the melting curve has multiple domains and the sequence change affects only one of them (Wittwer et al., 2003). However, in many cases, the shape of the homozygous variant curve is nearly identical to the wild-type curve. Therefore, as a general rule, temperature shifting is not performed when the objective is to identify homozygous changes and absolute temperatures must be relied upon. This places stringent requirements on the temperature precision of an instrument between samples, either in time (HR-1™) or space (LightScanner™). In practice, the HR-1™ performs better than the LightScanner™ because of temperature variation across sample wells in heating blocks. Fewer homozygous variants will be detected on 96- or 384-well instruments than on the HR-1™.
Systematic studies to estimate the sensitivity and specificity of homozy-gous variant detection have yet to be performed. Detection of homozygous SNP changes is often easier than small, homozygous insertions or deletions because the later may be very close in Tm to the wild type. For example, the homozygous G542X SNP in cystic fibrosis is detectable by high-resolution melting, while homozygous F508del is not (Chou et al., 2005). Studies with small amplicons have shown that 84% of human SNPs have homozygous Tm differences between 0.8-1.4°C (Liew et al., 2004), suggesting that most SNPs are detectable. Whether they are detected will depend on the ampli-con size and the resolution of the instrument used.
Was this article helpful?