This article was originally published March 18.
As next-generation sequencing technology continues to advance, bottlenecks are increasingly at the front and back ends of the sequencing workflow, with researchers looking to sequence genomes with smaller amounts of input DNA and in increasingly faster times.
Companies like Agilent and Illumina are aiming to address those needs with new exome kits. Illumina recently began shipping its Nextera rapid exome kit at the end of the first quarter and Agilent recently launched its HaloPlex exome kit for a faster turnaround time.
One early-access customer of Agilent's HaloPlex exome kit, Wilfred van IJcken, head of Erasmus Medical Center's Center for Biomics in the Netherlands, told In Sequence that he began testing it at the end of 2012. He said that the performance was comparable to Agilent's SureSelect kit but it enabled him to start with less input DNA and reduced the capture step to three days from six.
"It's always difficult to compare different techniques," he said, "but the vast majority of SNPs are concordant."
Van IJcken said that while he has not yet compared HaloPlex to kits from other companies, he plans to try Nextera's rapid exome capture kit when it becomes available.
He said that he plans to use the HaloPlex kit for both research and clinical projects for which there is limited starting material. "We don't always have the several micrograms that are needed" for SureSelect and other kits, he said. "That was a big reason for us wanting to try [the Haloplex] method out."
The HaloPlex kit enables as little as 200 nanograms of starting DNA in part because it uses restriction enzymes to digest the genomic DNA, which is an efficient process, he said.
By contrast, SureSelect uses shearing to digest the DNA, which results in some fragments that are too short and some that are too long, thus losing starting material.
With HaloPlex, there is also no need for purification following digestion, end repair, or A-tailing, steps that all result in DNA loss.
HaloPlex makes use of partially double-stranded probes to "directly fish for the fragments that you want to sequence," van IJcken added. It fishes for both ends of the fragment, allowing it to select for fragments that differ from the reference sequence, while SureSelect probes fish for a 120 nucleotide continuous fragment, making it more difficult to capture fragments that differ from the reference.
Additionally, the HaloPlex process has fewer total steps than the SureSelect method.
Van IJcken discussed his experience with the kit at an Agilent-sponsored workshop at last month's Advances in Genome Biology and Technology in Marco Island, Fla. He tested it on 12 samples — 10 of which were run on two lanes of the HiSeq 2000, one sample run on the MiSeq, and one sample run on the HiSeq 2500.
For each sample, he started with 200 nanograms of input DNA and used eight pairs of restriction enzymes to generate 2 million restriction enzyme fragments.
For the HiSeq samples, he generated an average of 50 million reads per sample, 97 percent of which aligned. Around 8 percent of the reads were lost in the "restriction fragment trimming" process while the MiSeq data had an even bigger trimming loss of around 31 percent, he said.
Comparing the data to SureSelect, "HaloPlex outperforms SureSelect in coverage per gigabase pair," he said, with a mean coverage of 51-fold to 57-fold for HaloPlex compared to 40-fold to 48-fold for SureSelect. HaloPlex had slightly less coverage of the first exon, but the mean coverage made up for the discrepancy, he said.
HaloPlex also missed slightly more bases, around 1.1 megabase pairs compared to 0.8 megabase pairs that were missed by SureSelect.
However, the missed bases were more randomly distributed in SureSelect than HaloPlex, said van IJcken. "In HaloPlex, you have certain exons that seem to be missed completely," he said.
The two methods were pretty equal when it came to SNP calling, with a concordance of 99.6 percent. Out of 35,000 total SNPs called, there were 52 SNPs called heterozygous by HaloPlex and homozygous by SureSelect, and 97 called homozygous by HaloPlex and heterozygous by SureSelect.
The HaloPlex method also displayed slightly less reference bias than SureSelect and was a bit better at detecting indels. For example, HaloPlex correctly aligned a known deletion that was misaligned in SureSelect, resulting in a false variant call.
Van IJcken attributed HaloPlex's better performance with indel calling to the fact that it uses restriction enzymes to digest the genomic DNA.
"All the reads you get from a certain restriction fragment are starting from the same point," he said, and "the indel is exactly aligned the same way in that fragment."
This is in contrast to SureSelect, where "reads are starting from different start points in the genomic location" so, for instance, if there is a mismatch in the last few bases of an alignment, the read could align to a location where there is an insertion or deletion. Then, it "gives some coverage to that region, which should actually be placed to another location after the insertion or deletion," he said.
Moving forward, van IJcken said that one important factor for customers to adopt HaloPlex would be automation. Agilent currently has automation available for SureSelect that is not yet amenable to HaloPlex, so for sequencing a large number of exomes, SureSelect still has an advantage, he said.
Another factor will be how the price of HaloPlex compares to SureSelect and other exome capture kits. The company has not yet disclosed pricing for HaloPlex.