NEW YORK (GenomeWeb) — Researchers from Qiagen have described the amplicon sequencing method that the company uses in its GeneRead panels and how it avoids potential "blind spots" that may be present in other amplicon sequencing methods.
The researchers described the blind spot problem and how to avoid it in a paper published this month in BMC Genomics.
Essentially, the group showed that when doing amplicon sequencing there are a couple of bases immediately adjacent to the amplicon primer where variants are likely to be missed due to a "blind spot" caused by misalignments.
In their study, the Qiagen researchers first showed that this issue was real and then demonstrated a solution to the problem.
One known issue of next-gen sequencing is that toward the ends of reads, alignments tend to be more inaccurate and if there is a variant at the end of the read, it can lead to a false negative call or even a false positive call.
"Many people that work in the field are aware that the alignments can be inaccurate near the ends of the read; that is not specific to amplicon reads," lead author Ravi Vijaya Satya told GenomeWeb.
In sequencing applications where read fragments are randomly generated and thus have different start and end positions, sequencing to enough depth will resolve the issue. However, with amplicon sequencing, the 5' end is defined and when enzymatic primer digestion is done, the 3' end is also defined.
"I'm not sure that everyone realizes that when it comes to amplicon sequencing, this can actually lead to a problem" with variant calling, Satya said.
Enzymatic primer digestion is done by first substituting one or two thymine bases in the primer with uracil. During PCR, adenine bases will be incorporated into the complementary positions with uracil in the primer. Treating the DNA product with uracil-N-glycosylase will then remove the primer up through the furthest uracil base leaving a single-stranded overhang. That overhang can then be clipped with an exonuclease reaction.
In the study, the researchers demonstrated that enzymatic primer digestion of amplicons prior to sequencing can lead to false negative calls or false positive calls near the amplicon junction, and also demonstrated techniques for resolving the issue.
The researchers targeted coding regions of 167 cancer genes with 8,035 primer pairs. The amplicons ranged in size from 100 bp to 180 bp and primers were between 17 bp and 28 bp. Then for each amplicon, the team generated a point mutation at the first base after the primer, generating more than 15,000 point mutations.
Next, they created two haplotype sequences from each amplicon by incorporating just one mutation into each haplotype, and generated paired end sequence reads at 100x depth from each haplotype.
Amplicons overlapped such that any given base would be covered by between one and three amplicons. They also performed primer digestion, analyzed the results, and found that the closer primer digestion was to the actual amplicon, the worse the sensitivity was for calling the simulated variant.
For instance, when primer digestion occurred at a median distance of just one base away from the amplicon, only 82.17 percent of the variants were called with 2,830 false negatives. Sensitivity increased to 97.5 percent when the median distance increased to three bases, but there were still 397 false negatives. It wasn't until the researchers increased median distance to 10 bases that they achieved 100 percent sensitivity with no false negatives.
"Our analysis suggests that primer bases should be part of read alignments and subsequent post-processing of the alignments to ensure variant calling with high sensitivity," the authors wrote. "If enzymatic primer digestion is used, we argue that at least a few primer bases should be left undigested."
Satya said that researchers can employ a number of different strategies to avoid the problem. One solution is to design amplicons in such a way that there is a lot of overlap.
"That kind of design will require more amplicons to cover the target region to make sure you have an overlap between adjacent amplicons," which will require a higher level of multiplexing, he said. "But, it should be possible to avoid the blind spots even if you are using enzymatic primer digestion."
For Qiagen's GeneRead panels, Satya said that the firm has moved away from using enzymatic primer digestion. Instead, he said, the firm "includes the primers in the alignment," removing them "just before doing the variant calling without altering the rest of the alignment."
A third option, he said, is to still do enzymatic primer digestion, but to leave a few primer bases after enzymatic primer digestion.
Satya said that in Qiagen's process of designing its GeneRead panels, "we weighed all the pros and cons and we went with not using enzymatic primer digestion at all."
This helps the problem, he said, because any misalignment toward the end of a read is contained within the primer sequence rather than the amplicon itself. "We're avoiding those regions, what we call blind spots," he said.
Satya said that although the group did not test specific commercially available amplicon sequencing, "we expect to run into this problem if enzymatic primer digestion is applied" with the exception of hotspot panels.
For instance, a number of commercial cancer hotspot panels use amplicon sequencing to look at specific locations within genes that are known to be frequently mutated in cancer.
Hotspot panels are usually designed in such a way that the variant location is at the middle of the amplicon, so the accuracy of calling that variant wouldn't be affected by the inaccuracies toward the end of the amplicon, Satya said. But, a gene panel that is looking at all the coding regions in a gene would have that problem, unless steps were taken to get around it, he added.
Satya said that Qiagen has been using this method in its commercial GeneRead panels and plans to continue using it for future products. He added that the goal of the paper was to "bring awareness to this issue that people could be running into without realizing it."