By Julia Karow
Sequencing platforms from Illumina and Helicos BioSciences each offer unique features for analyzing ancient DNA, and scientists should pick the right tools depending on the degree of damage and the biological question they want to answer, according to researchers in Denmark.
In an effort to explore different technical approaches for sequencing the genomes of ancient horses, Ludovic Orlando and colleagues at the Centre for GeoGenetics at the Natural History Museum of Denmark recently compared the Helicos HeliScope and the Illumina GAIIx platforms, publishing their results online in Genome Research last month. They also plan to test other new platforms for ancient DNA sequencing, including the PacBio RS and the Ion Torrent PGM.
"The main idea was really to test and develop methods to improve paleogenomics, the characterization of ancient genomes," said Orlando, an associate professor at Copenhagen University. Specifically, his group is interested in sequencing the complete genomes of ancient horses to compare them with the genomes of domesticated horses. Before embarking on these projects, though, he and his colleagues decided to spend several months exploring different technologies and methods.
Based on the results, they chose to sequence one horse bone at low depth using a combination of Helicos and Illumina, and another, younger bone with high coverage using the Illumina HiSeq platform. The researchers are currently analyzing data from these projects and plan to submit their results for publication within the next six months or so.
Ancient DNA tends to be fragmented and damaged, and is contaminated with large amounts of microbial DNA. Because sample preparation for the single-molecule Helicos sequencer is much simpler than for the Illumina and does not involve amplification, the researchers thought that more of the precious ancient molecules might be sequenced on that platform. "[We wanted to] test if building libraries [on the Illumina platform] is detrimental to accessing the true diversity of the molecules present," Orlando explained.
For their comparison, they sequenced DNA from a permafrost-preserved Pleistocene bone that is at least 50,000 years old, generating 116 megabases of data on the HeliScope and 77 megabases on the Illumina GAIIx. Helicos performed the sequencing at its facility, while the Danish National DNA High-Throughput Sequencing Centre, a core facility housed in the same building as the Centre for GeoGenetics, performed the Illumina sequencing.
The percentage of horse DNA sequenced by the Helicos platform was indeed higher than for the Illumina sequencer, they found, a small but significant difference. According to Orlando, some of the short, fragmented molecules might be refractory to the enzymes used during the Illumina library prep, and the amplification step might favor some molecules over others.
The finding that Illumina libraries are biased towards microbial sequences over ancient DNA was "the most surprising aspect" of the paper to him, according to Meyer. "This raises hopes that improvements in library preparation could lead to a substantial reduction of sequencing costs for paleogenomes," he said.
The Danish researchers further increased the number of horse DNA molecules sequenced on the Helicos platform by about three-fold by lowering the temperature of the DNA denaturation step. Additional improvements to the Helicos sample preparation and image analysis process that increase the sequencing signal and the number of accessible ancient DNA molecules will be reported in an upcoming paper, Orlando said.
In addition to sequencing a greater percentage of horse DNA, the Helicos platform also helped discover a new class of DNA damage in ancient DNA that results in single-stranded 3' overhangs. "This was impossible to detect with the Illumina technology," he said.
Besides comparing sequencing technologies, the researchers also improved the process of extracting DNA from bones. They found that DNA from a second digestion yielded more ancient DNA molecules than the first extract. One possible explanation, he said, is the existence of "molecular niches," such as large crystals, that favor the preservation of ancient DNA and take more time to digest than other parts of the bone.
[ pagebreak ]
The Right Tool for the Task
Following their technical evaluation, the researchers chose to sequence the Pleistocene horse at low coverage with a combination of Helicos and Illumina, probably to one-fold depth for each platform. "Of course it takes a little more time, and more money as well, but I think the quality of the draft that we have now was really worth it going with the two technologies," Orlando said.
While the Helicos technology provided a higher percentage of horse reads, it generates shorter reads than Illumina — about 25 to 50 base pairs in length compared to 75 base pairs or more. Also, since Helicos has never before been used to sequence ancient DNA, the researchers wanted to be able to compare the Helicos draft genome to one generated with a well-characterized technology. Finally, uncertainty about Helicos' future necessitated that the team had a backup option. As of June 30, the firm had $1.6 million in cash and said recently that it plans to reduce its headcount to 10 by the end of September (IS 8/16/2011).
At the same time, the scientists decided to sequence a 13,000-year-old horse bone at high coverage using only Illumina's HiSeq, which has "amazing" throughput, he said. Because that specimen is younger than the other one, its DNA is less damaged, and thus it has been faster and easier to use Illumina only, especially in the face of competition from other groups.
When choosing a platform for sequencing ancient DNA, Orlando said, researchers should be guided by both the quality of the DNA and by the biological question at hand. By and large, he said, the Helicos platform has advantages for sequencing highly damaged DNA because it analyzes a larger percentage of the molecules and because its short reads match the short DNA fragments.
When there is less damage and longer reads would be useful, and when time is of the essence, however, the Illumina HiSeq could be a better choice. "If you have to go fast, the throughput of HiSeq is such that it outperforms any kind of platform that is available," he said.
The Helicos platform could also be useful for targeted sequencing because it can capture regions of interest directly on the chip, he said, although he and his colleagues have not yet tested this. "If you are interested in surveying the diversity of the exome of an ancient species, then this might be the right way to go," he said. Helicos recently demonstrated the direct capture method for sequencing the BRCA1 gene (CSN 8/9/2011).
"Applying new sequencing technologies to ancient samples is very exciting, as this may advance our understanding of types and extent of chemical modifications in ancient DNA," said Matthias Meyer, a postdoc in Svante Pääbo's group at the Max Planck Institue of Evolutionary Anthropology in Leipzig, who was not involved in the work. That knowledge, he said, is not only of academic interest but could help develop new methods to obtain more information from the molecules.
But according to Meyer, complete ancient genomes are unlikely to be sequenced with Helicos technology alone anytime soon because its error rates are high and "the reduced bias in sample preparation comes at the expense of complex biases in sequencing."
He said he believes, though, that single-molecule sequencing in general "will have a big impact on ancient DNA research." Nanopore sequencing in particular seems interesting, he added, because it may be able to identify modified DNA bases directly. The paper, he said, "provided a first glimpse of what sequences from ancient samples may look like in the future."
Orlando and colleagues are also keeping an eye on emerging sequencing technologies and plan to test both the single-molecule Pacific Biosciences and the Ion Torrent platforms in the coming months for ancient DNA sequencing.
[ pagebreak ]
The allure of the Ion Torrent PGM is its short run time, he said, since its throughput is currently much smaller than that of the HiSeq or Helicos. Prior to large-scale projects, many samples need to be screened in order to select one for whole-genome sequencing, and "in that phase of screening, it might be really interesting to use [the Ion Torrent] platform," Orlando said.
On the PacBio, the circular consensus sequencing mode, where the same molecule is sequenced several times over, could be useful for ancient DNA analysis. A single run, though it only produces a fraction of the reads of the Helicos or Illumina platforms, is very quick and costs a few hundred rather than thousands of dollars, he said. Also, he and his colleagues simply need to stay ahead of the curve. If a new technology becomes the dominant platform in a few years, "you have no choice, you have to use it," he said. "We have to anticipate that."
Meyer said that researchers at the Max Planck Institute have also been using the Ion Torrent PGM, though the workhorse for most projects is still the Illumina platform. "It is generally very challenging to implement new technologies," he said, because tools developed for one platform do not necessarily work on others, and because each platform has its own challenges.
Have topics you'd like to see covered in In Sequence? Contact the editor at jkarow [at] genomeweb [.] com.