Jeffrey Perkel has been a scientific writer and editor since 2000. He holds a PhD in Cell and Molecular Biology from the University of Pennsylvania, and did postdoctoral work at the University of Pennsylvania and at Harvard Medical School.
When it comes to next-generation DNA sequencing, researchers can pretty quickly sequence an entire genome, but they don’t have to, and often, prefer not to.
Maybe they’re interested only in specific genes or genomic regions, for instance. Or perhaps funding is limited. Whatever the reason, many researchers elect to focus on targeted genome subsets, ranging from complete protein-coding exomes to smaller, more targeted panels.
There are two predominant strategies available for DNA target enrichment. Agilent Technologies’ SureSelect panels, Roche NimbleGen’s SeqCap EZ panels, Illumina’s Nextera® Rapid Capture Exome panel and Integrated DNA Technologies’ xGen® Lockdown® panels all rely on solution hybridization, in which pools of biotinylated oligonucleotide probes, each targeting a segment of the desired capture region, are used to pull down particular subsets of the genome.
RainDance Technologies’ ThunderStorm™ platform and Zymo Research’s targeted bisulfite sequencing service (for enrichment of selected methylated DNA regions) depend on PCR enrichment, in which each targeted region is selected via a distinct set of primers in a kind of massively multiplexed reaction. RainDance separates each pair of primers in picoliter-sized droplets, effectively creating millions of individual reaction vessels that each carry out a single amplification reaction. Zymo uses Fluidigm’s 48.48 Access Array™ to set up as many as 48 different PCR reactions for each of 48 samples, according to Xuegang Sun, Zymo’s director of platform technology.
Other possibilities include hybridization microarrays and molecular inversion, or “padlock” probes, a form of PCR multiplexing.
Whichever platform you choose, a few off-the-shelf “catalog” enrichment panels typically are available. Companies develop catalog designs they think will sell, of course. So exome panels and panels for widely studied human diseases or gene classes are relatively common while those targeting niche biological processes and model organisms are not. Agilent Technologies, for instance, offers a premade panel for its SureSelect system targeting the so-called kinome, which comprises the 500-plus kinase genes encoded in the human genome. RainDance’s ThunderBolts™ Cancer Panel targets 50 oncology-related genes, while Roche NimbleGen has developed panels targeting cancer-related genes (578 genes) and genes implicated in neurological disorders (256 genes).
Often, though, users’ needs don’t precisely match the content of catalog designs. Maybe those premade targets are close but are missing a few key genes. Or maybe the user is interested in less well-studied biological processes or model organisms. In such cases, custom enrichment designs can help.
Balancing throughput and cost
Designing a custom panel is relatively straightforward. Basically, supply a list of genes or genomic addresses (for instance, using Agilent’s SureDesign or Roche NimbleGen’s NimbleDesign software), and the company does the rest. (A recent paper in Nature Protocols details the process [1].) But there are several variables to consider.
First, how much genomic sequence are you actually targeting? A typical exome targets about 50 Mb of DNA, but most custom designs target far less, “maybe less than 1 Mb to a few megabases in size,” says Yong Yi, marketing director for next-generation DNA sequencing and gene regulation at Agilent Technologies. As a general rule, the larger the target area, the higher the cost, as larger targets require more probes to be synthesized and greater optimization.
Next, how many samples do you need to process? It generally makes no sense to build a custom panel for a handful of samples, as custom designs take time to develop and can be relatively expensive, says Mark Behlke, chief scientific officer at Integrated DNA Technologies (IDT). Premade panels, in contrast, are “game-changers,” he says, and IDT offers three of them, a pan-cancer panel targeting 127 genes, an acute myeloid leukemia panel (260 genes) and an inherited diseases panel (4,503 genes). “The customer doesn’t have to buy the whole lot, they just get an aliquot and thereby get access to a large, costly panel at a low price point.” IDT sells custom panels for $15 to $35 per bait, depending on desired oligo yield, Behlke says.
IDT synthesizes and QCs each oligo individually, Behlke notes, adding that while that strategy can be more expensive up front, actual cost per sample can be lower than other approaches when studies involve hundred to thousands of samples since the yield obtained on custom syntheses is so high. Plus, using individual synthesis means each oligo can be individually validated for sequence and concentration, thereby ensuring that each bait is correct and present in the final pool at the same concentration.
S. Roopom Banerjee, president and CEO of RainDance Technologies, says his company’s custom designs can accommodate up to 20,000 targets. With an average of 10 targets per gene, that means a fully loaded RainDance panel can target some 2,000 genes. However, 95% of customers use fewer than 3,000 targets, Banerjee says. The user simply supplies a list of gene targets, and the company designs and “dropletizes” PCR primers for each.
Three thousand targets require 6,000 custom primers, and though oligo prices have fallen sharply over the years, that's still a lot of bases to buy. The typical panel costs $10,000 to $15,000, Banerjee says, and very large panels can cost up to $100,000. It can be hard to justify that expense for a panel that will be used just a few times. But the company supplies enough material for about 2,000 samples, Banerjee says, so if you have a locked-down design and anticipate processing large numbers of samples, “it’s pretty cost-effective for that small upfront investment.”
Another variable to consider is overlap with existing designs. Do you need a subset of a predesigned panel? It may be possible to acquire only those probes, which have the benefit of previous validation. Alternatively, can you get by if you simply add, say, 20 targets to an existing catalog design? Agilent, at least, allows users to make those changes via its SureSelect Plus option, Yi says.
Finally, consider the overall complexity of your desired target. It’s relatively easy to come up with probes or primers targeting unique genes, but what about large gene families and pseudogenes? Depending on the complexity of the design and the user’s goals, it can take a while to design and optimize panels to, for instance, pull out one member of a family to the exclusion of its siblings. “It’s an open, ongoing discussion [between the company and the user] to find out what you want before we pull the trigger and perform the synthesis,” Behlke says.
But before committing, Yi recommends users request the "design file," which details the genomic locations the probe set targets. This can then be uploaded to a genome browser to ensure all key regions are covered. Design algorithms often filter out repetitive regions, for instance, Yi explains, and this step can give users the chance to make changes.
Just exome it?
Another possibility is to sequence the exome to begin with—after all, those panels are prepared, optimized and ready to go, and the panels can be less expensive than custom designs (at least for small numbers of samples).
One stream of approaches, says Ji Wu, international marketing director at Roche NimbleGen, researchers are opting to sequence the entire exome and then focus on particular target genes during downstream data analysis. “The beauty of that is if you want to look at 500 genes initially and you don’t find what you need, you can always go back and do a broader data analysis,” Wu says.
Yet even today, bioinformatics remains a bottleneck, says Yi, and the more DNA being targeted, the more complicated the analysis. “Having a smaller custom [design] makes analysis easier.”
Plus, given finite sequencing capacities in even the most high-end sequencers, larger targets inevitably translate into fewer samples per run and/or shallower sequencing coverage, which increases costs. At least in oncology, Wu says, “people want an average of 500-fold coverage when they have a very focused gene panel.”
Ultimately, says Wu, a custom design is, well, custom, and depends on users’ needs. What depth of coverage do they require? What sorts of samples are they using, and what types of variants are they looking at? “There’s a lot of things you need to consider,” she says.
Reference
[1] Mercer, T.R., et al., "Targeted sequencing for gene discovery and quantification using RNA CaptureSeq," Nat Protoc, 9:989–1009, 2014. [PubMed ID: 24705597]