About a year after first presenting the Polonator sequencer to the scientific community and shipping instruments to early-access users, Dover, a Danaher Motion company, is now ready to sell an upgraded version of the instrument to a wide range of customers.
A few weeks ago, the company, which has been co-developing the low-cost, non-proprietary, high-throughput sequencing platform with George Church's group at Harvard Medical School, shipped an instrument to the University of Utah, its fifth customer to date and the first lab outside the early-access circle of users.
Over the next year or so, the developers plan to make additional improvements, including reagent kits, increased read lengths, and a simpler library preparation protocol. By 2010, they expect the output per run to increase tenfold, to 100 gigabases.
Several early-access customers have used the Polonator for bacterial genome sequencing and other projects and are planning to use it for a variety of sequencing applications in the future.
The recent upgrades, which include changes to the flow cell, fluidics, operating software, and instrument cover, resulted from feedback from early-access customers, according to Kevin McCarthy, chief technology officer of Dover. "It took longer than we expected to address the issues raised in our early-adopter phase, but all changes have been implemented, and we now have a robust, high-performance sequencer," he told In Sequence.
Without bead enrichment, the instrument currently generates more than 10 million mappable reads per flow cell lane, or about 5 gigabases of data per run, based on 2x13-base gapped paired-end reads and two 8-lane flow cells. Bead enrichment is expected to double the output per run to 10 gigabases, according to McCarthy.
Consumables costs for generating a megabase of data, or 40,000 sequence tags, are currently $1, and are expected to drop approximately tenfold over the next year, he said.
The instrument itself now costs $170,000, which is $20,000 more than a year ago (see In Sequence 2/5/2008), due to price increases for a number of components, including the camera. Set-up costs are not included in this price.
Reagent kits are not yet available but Dover plans to offer these at a later stage when read length increases. In the meantime, customers can purchase their own reagents, which are all non-proprietary.
In addition, one of the early-access customers, Jeremy Edwards at the University of New Mexico, is offering a library construction service for Polonator users.
Further information about the instrument is available on the Polonator website, which will be updated with additional data over the next few weeks, McCarthy said.
He said the instrument should appeal to users who keep an eye on cost as well as those interested in testing new sequencing chemistries. Users "can get in the driver's seat and take complete control of the instrument with very little effort and use it to innovate," he said. "We believe that even though they would be a small fraction of the user base, their contributions could be dramatic."
According to Rich Terry, a senior research engineer who leads the Polonator development at Harvard Medical School, the instrument is "extremely fast," both in terms of biochemistry and image acquisition. In addition, its cost is low compared to other sequencing platforms and it has applications beyond sequencing. "For the cost of a fully-automated Nikon microscope, you could get a fully operational DNA sequencer that can also do all of your imaging for you that you would do on a standard microscope," he said.
[ pagebreak ]
Dover may offer less technical user support than the manufacturers of competing sequencing platforms, but McCarthy does not expect this to be a problem. The instrument is very reliable, he claimed, and "we expect a growing user community to quickly develop who will share expertise" on how to use the platform.
One major upgrade to the early-access instrument has been new software, which now has graphical user interfaces and is more robust and easier to use. It also supports different running modes, and allows users to change protocols according to their needs, for example to test different sequencing chemistries.
"It makes it really user-friendly at a research level, so it can be used pretty much for anything you want in terms of microscopy and scanning applications," said Terry.
In addition to changes in the fluidics system, the flow cell now has 8 lanes — like the flow cells Illumina uses for its Genome Analyzer — instead of 18 circular wells. This redesign, among other things, allows reagent volumes to be smaller.
Finally, the outside cover or "skin" of the instrument now consists of three pieces instead of one, making it easier to access the machine's interior.
The changes did not only address problems with the instrument, but were also made with an eye towards future improvements. "Any of the changes we look to make in the future we already built into the existing device, so you can get them as simple upgrades, either software or a small bolt-on piece of hardware," Terry said.
All early-access Polonators — three at the Church lab at Harvard, one at the University of New Mexico, one at the Max Planck Institute for Molecular Genetics in Berlin, and one at the Broad Institute — have either been upgraded already or are in the process of being upgraded.
During the six months that his lab has had the Polonator, Jeremy Edwards, an assistant professor of molecular genetics and microbiology at the Cancer Research and Treatment Center of the University of New Mexico Health Sciences Center, has used the instrument to sequence about 60 bacterial genomes.
One of the reasons he said he acquired the Polonator is that part of his research is devoted to technology development, "and obviously, the Polonator allows for that."
In addition, the low cost of the instrument and consumables was "one of the biggest reasons" for his decision, said Edwards, a former postdoc in Church's lab. At the time he investigated different platforms, he estimated the Polonator was five times less expensive to run than competing platforms, although he acknowledged that the cost has since come down on other systems as well.
Much of his team's effort so far has been devoted to optimizing the library protocols, because while the sequencing worked well, the coverage of the genomes was not as expected. "I came to the conclusion very quickly that the Polonator is working great, but we needed to feed it with better input materials," he said.
Having spent a lot of time on library prep, Edwards is now planning to offer library construction services to Polonator users. Initially, these will be genome sequencing libraries, but his lab is also working on protocols for transcriptome sequencing, as well as ChIP-seq and other tag-based protocols.
His lab is also working on increasing the read length to 2x24-base reads, although "we don't have that completely nailed yet," he said. But even paired-end reads as short as 2x15 bases would be useful for applications such as human genome sequencing, he reckons. "Anything beyond that is just icing on the cake," he said. "The key is, you really need to have mate pairs."
Eventually, he wants to use the Polonator to sequence human genomes, in particular cancer genomes.
[ pagebreak ]
Edwards said his original perception that the Polonator was going to appeal only to a small group of technology-savvy users was probably wrong. He is currently considering transferring his original instrument to a genomics core facility at UNM and to acquire at least one more instrument for his own research.
"It's quite easy to use, it doesn't require all the knowledge of coding, and it can be run from a graphical user interface in a very easy way, with the added advantage of having access to all the source and having the details of everything that's happening internally," he said. "I think it could be applicable to many people, with the caveat of somebody who needs really long reads," which the Polonator does not currently provide, he added.
Regarding data-analysis software, he said that existing software is likely to have problems with the current, gapped reads. As a result, he has developed his own software for mapping and statistical analysis, which he plans to make available to users upon publication.
Also, once the reads have been improved and no longer have gaps, he said, existing mapping software "will work just fine."
Kael Fischer, a research assistant professor in the department of pathology at the University of Utah, just received his Polonator a few weeks ago and had not yet run it as of last week.
The main reason he chose the platform was its "very attractive" price, compared to other sequencing platforms that allow for paired-end sequencing. At the time he made his decision at the end of last year, the paired-end module for the Illumina Genome Analyzer alone would have cost him $200,000, more than the Polonator, he said.
Based on data provided by Dover and current Polonator users, and how much two Illumina GA users at the University of Utah paid for consumables at the end of 2008, he also calculated that the Polonator would provide sequence tags more cheaply. "It doesn't come out much cheaper in a per-gigabase comparison, but in a per-tag comparison, it really is a very significant savings," he said, an advantage for tag-based applications like gene-expression analysis and ChIP-seq.
Fischer also compared the cost of sequencing to that of microarrays, which he wants to replace with the Polonator in pathogen discovery experiments. When he was a staff scientist in Joe DeRisi's group at the University of California, San Francisco, he and his colleagues used home-spotted microarrays at a cost of approximately $20, and he said that with the new sequencer, "I think we are pretty close to that."
"The Polonator was the first time that I saw the cost of sequencing a sample approaching the cost of doing a microarray analysis," he said.
Fischer plans to use the new instrument for pathogen discovery from clinical material, and to apply it more widely in the pathology department as a tool for gene expression and ChIP analyses.
He does expect some growing pains, as they have occurred with other new platforms. "All the companies have problems in the beginning," he said. "My experience with Dover is that they — and this has probably something to do with why this release has taken so long — are very careful with what they are releasing. They really don't want to release something that is going to be a waste of your time and your precious samples."
Many new developments, he believes, will likely come from the user side. "I think that the Polonator has the potential, as the community grows, to really be a very rich platform," he said. "It's really just a fluidics package with a camera. And then the chemistries, the software, and the analysis system, all of that is as good as the community makes it."
In the meantime, Dover, the Church lab, and the Edwards lab are working on a variety of improvements to the platform and protocols. "We believe we will maintain at least parity with [other next-gen players] in terms of throughput, we will have sufficient read length, and we will have a significant cost advantage in the instrument and in the consumables and reagents," McCarthy said.
[ pagebreak ]
Barcoding protocols are one area of development, with the goal of running 256 samples per run. The Church group is also working on improving the sample preparation, aiming to replace the cumbersome emulsion-PCR with rolling circle amplification colonies, or "rollonies," which do not involve beads and which are somewhat similar to the nanoballs Complete Genomics uses, according to Terry.
Paired 24-base reads "are showing good results," McCarthy said, and are expected to be available later this year. Also, Terry and colleagues are exploring several ligation-based, polymerase-based and other sequencing chemistries, developed at Harvard and elsewhere, that could potentially further increase the read length.
In addition, in order to improve the read density, the developers expect to transition to ordered arrays later this year.
Sometime in 2010, they aim to reach an output to 100 gigabases per run, according to McCarthy. This goal is in line with that of Illumina and Applied Biosystems, who both revealed roadmaps earlier this year for getting to 100-gigabase runs for their Genome Analyzer and SOLiD platforms by the end of this year (see In Sequence 2/10/2009).