NEW YORK (GenomeWeb) – Scientists from Bio-Rad Laboratories and Harvard Medical School have reappropriated droplet digital PCR (ddPCR) technology to detect genomic sequence variations linked on the same chromosome.
The scientists believe the method of determining chromosomal phase, which they called Drop-Phase, could lead to more haplotype analysis in genomic studies of disease and should garner interest from clinical genetics labs.
"It's a conceptual leap from using ddPCR to count things to using it to determine what's physically attached to what," Steven McCarroll, a faculty member at the Harvard Medical School and a senior author on the paper, told GenomeWeb.
Drop-Phase is a fully automated technique to quickly determine whether pairs of specific alleles are located on the same chromosome in the genome. The workflow is essentially identical to a normal ddPCR workflow, without a step for restriction digestion.
In each Drop-Phase reaction, genomic DNA segments are isolated in tens of thousands of nanoliter-sized droplets together with allele-specific fluorescence probes, in a single reaction well, the authors wrote in a paper describing the technique and published in March in PLOS One. Physically linked alleles are found in the same droplets and the increase in fluorescence compared to background noise can be detected by Bio-Rad's proprietary QuantaSoft software.
The researchers were able to phase CFTR alleles as far apart as 116 kilobases in the genomes of cell lines derived from cystic fibrosis patients.
"Determining chromosomal phase of multiple sequence variants is an old and classic unsolved problem in human genetics," McCarroll said. "It's a problem that we confront all the time in our work." Finding out if two SNPs are linked on the same chromosome is possible using techniques such as long-range PCR, but becomes more difficult the further apart the SNPs are. "On the scale of most genes, especially genes in nervous system, the distance is longer than the scale of PCR amplicons," McCarroll said.
Jack Regan, a scientist at Bio-Rad and co-lead author of the paper with Harvard's Nolan Kamitaki, added that without good solutions to phasing variants, making progress in haplotype analysis of genomes has suffered. Outside of long-range PCR and whole-genome sequencing, "there were no good solutions," he said.
"We expect researchers to use this technique to phase variants discovered from sequencing; and to phase GWAS samples to determine whether a particular haplotype has clinical significance," Regan said. McCarroll predicted that there would be "substantial interest" in the technique from clinical genetics labs.
Neighboring genetic variants can affect the expression levels of other genes as well as protein folding and binding sites. "As we dive deeper into the genomics of diseases, these biomarkers that are complex due to haplotype are going to become much more common," Regan said.
Genetic variation of the CFTR gene linked to cystic fibrosis provided a good model because disease severity can change depending on whether variants are located on the same chromosome or different ones. Two of the samples were confirmed to have a complex allele comprised of two mild variants that were configured on the same chromosome to form a pathogenic haplotype. "This is a great example of how variants can have a different phenotype if they are found to be cis-configured," Regan said.
The main difference between running ddPCR and Drop-Phase was that the researchers omitted the restriction digestion step of sample preparation. Other than that, the study was contained in the normal workflow, although there were several extra considerations, Regan said.
"The quality of DNA going in is very important in terms of how far out you can assess linkage," Regan said. "If the targets are far apart, greater than 40 kilobases, you need to be careful with the extraction chemistry to preserve the integrity of DNA," he said. "We focused on automatable chemistries since we believe this technology will be used for population-based studies that involve hundreds or thousands of samples." But he added that non-automated extraction chemistry could potentially provide better results for more distant pairs of variants beyond 210 kilobases, the maximum distance demonstrated in the paper.
There are a few limitations to the technique. First, each set of alleles has to be known to be heterozygous. "If you have one locus which is homozygous, where it's the same on the paternal and maternal chromosomes, there's no phase information to be had," Regan said.
In the paper, the authors also noted that "though [Drop-Phase] scales quickly to large numbers of samples, it does not scale quickly to large numbers of loci, as each assay requires its own allele-specific fluorescence probes and optimization."
They added that genomic range is limited by sample preparation methods and reported that multi-well analysis did not substantially extend this range beyond the distances reported in the paper.
"It's important to understand how the phase of variants affects composition of proteins," Regan said, noting that complex diseases like Alzheimer's and schizophrenia involve many different genes and understanding the cause of those diseases is a great challenge. "Understanding how variants interact with each other at the gene level is very important," he said. "If you have two variations found in exons of proteins, that's going to affect folding of the protein, the binding sites."
"In some cases the biomarker is the haplotype," Regan said, "It's the combo of variants along the same strand of DNA. All of a sudden you can screen 1,000 samples for particular haplotype of interest, with low cost. This allows you to have better resolution genomically and allows you to better interpret your data."