By Julia Karow
This article has been updated to correct Tristan Orpin's title. He is chief commercial officer, not chief operating officer of Illumina.
As Illumina prepares to ship its MiSeq desktop sequencer more broadly, early-access customers are saying they are happy with their data and are considering using the instrument for a variety of applications that call for speed and moderate throughput.
Over the last few months, Illumina has run both a traditional early-access program for MiSeq, shipping instruments to more than 10 customers, and a customer sample program, under which it sequenced samples submitted by dozens of researchers.
"The number one reason we are doing the [customer sample] program, in alignment with the early-access program, was to improve our understanding of how people would use the platform," said Tristan Orpin, Illumina's senior vice president and chief commercial officer.
The company believes that MiSeq will not only be used in research, but also in diagnostics, forensics, and other applied areas, groups that it "did not have strong ties with in the past," Orpin said. Therefore, the goal of the program was to reach not only those users who are already familiar with next-generation sequencing but also customers who have so far stuck with Sanger sequencing, and even some without any direct sequencing experience, he said. "It's enabled us to understand them and their needs better."
Samples submitted by researchers — from the US, Asia, Europe, and the Middle East — ranged from viruses, bacteria, fruit flies, and snakes to human samples, many of the latter with clinical applications in mind.
Several projects focused on detecting somatic variants in panels of cancer genes, Orpin said, and a number of others targeted sets of genes involved in Mendelian diseases.
One customer, for example, sought to sequence Mendelian disease genes in human embryo cells prior to implantation, while another one, a hospital near Illumina's UK office, wanted to find the source of an outbreak of methicillin-resistant Staphylococcus aureus in a neonatal ward, a project that will be published.
The Broad Institute received its MiSeq instrument this summer and is using it "just about every day," said Chad Nusbaum, co-director of the Broad's genome sequencing and analysis program.
According to Nusbaum, the instrument is "pretty easy" to use, runs fast, and provides high-quality data, although at a greater cost per base than the HiSeq. It has been running according to Illumina's specifications, he said, and so far, there have been no serious problems with the machine.
According to Illumina, MiSeq produces more than 120 megabases of data with 35-base reads in four hours, and more than 1 gigabase of data with paired 150-base reads in 27 hours, including amplification and sequencing, and the number of unpaired reads exceeds 3.4 million.
The base accuracy of the data "is similar to what we see for the HiSeq," Nusbaum said. Toward the ends of the reads, the quality is even slightly higher than for HiSeq, probably because the sample spends less time on the machine.
Initially, the Broad plans to use the platform for "any kind of urgent project where turnaround time trumps cost of the data," Nusbaum said. This includes, for example, R&D projects, because "you get your answer in a day rather than in a week and a half."
In addition, projects that "fit nicely onto a small platform" will be run on MiSeq in the future at the Broad; these could include, for example, viral and microbial sequencing projects.
Generally speaking, because they generate similar data, applications for MiSeq and HiSeq would be more or less the same, Nusbaum said, but MiSeq offers greater speed and lower throughput. "The throughput of the HiSeq is now so big that it exceeds the need of some users," he said. And while MiSeq's yield is not high enough for routine human genome sequencing, it has the right scale for applications such as targeted or exome sequencing. "What you get is speed — if speed is part of our application requirements, then this is a good machine for that," he said.
[ pagebreak ]
Speed is one of the reasons the Cambridge Research Institute in the UK has been interested in MiSeq, according to James Hadfield, head of the genomics core facility at CRI. "We are very interested in a faster analysis of patient samples" for research purposes, he said.
After participating in the customer sample program, the institute now expects to receive its own MiSeq instrument later this month. CRI sent Illumina long-range PCR products of the p53 gene from a number of previously analyzed patient samples. The company processed these using the Nextera library prep method and sequenced them in three slightly differently configured runs, the latest run producing 5 million paired 76-base reads.
The data quality "is very good," Hadfield said, slightly better than data his lab has obtained from other Illumina sequencing platforms. "We are essentially seeing that the MiSeq generates the same data, and the same results, as a GA or a HiSeq."
Hadfield hopes the better data quality will allow Illumina to push the read length past 150 in the future, which could enable researchers to span entire small amplicons with paired reads.
Initially, he and his colleagues plan to use the MiSeq to check the quality of Illumina sequencing libraries prior to putting them on the HiSeq. The aim is to analyze up to 96 samples in parallel, generating as few as 100,000 reads per sample. "QC is going to be an important part of using MiSeq," he said. "Ultimately, it will make the results we get from genome sequencing better."
Methods development is another area where MiSeq will be used at CRI. Developing new sequencing protocols on existing next-gen sequencers "is laborious," Hadfield said, because the machines are often busy and a run takes up to 10 days. "MiSeq offers us the opportunity to … get a result after 24 hours, reanalyze that, going to the next step of the experiment, and very quickly develop those methods," he said. This will enable labs to test new methods more robustly before publishing them, and "will increase the quality of those methods publications," he said.
Finally, CRI wants to use MiSeq to sequence patient samples "where we want to get data quickly for research purposes," he said. Hadfield and colleagues have already combined Fluidigm's multiplex PCR Access Array system with Illumina's GA and HiSeq to sequence hundreds of loci in hundreds of individuals, and for some applications, "we will start to be able to run that on MiSeq and get a much quicker turnaround," he said. The plan is to develop mutation panels to screen incoming samples in order to decide which ones to analyze further by exome or whole-genome sequencing.
At the moment, MiSeq does not provide enough reads for ChIP-seq experiments, which typically require about 20 million reads, but that might change in the future if the number of reads per run goes up. Based on the imaging method currently used by Illumina, which he said only images one of the two surfaces of the flow cell, and not its entire area, "it certainly looks like that [the number of reads] could go up quite significantly," Hadfield said.
Like CRI, Graham Taylor's lab at the Leeds Institute of Molecular Medicine and Leeds Teaching Hospitals has tested the MiSeq by sending Illumina samples for sequencing, specifically short and long PCR products from tumor and non-tumor samples.
Illumina sequenced these samples in two runs, each yielding about 3 million paired 150-base reads with a read quality "equivalent" to the institute's own GAIIx platform. While the quality was "slightly lower" at the beginning of the read, it was very close and "perhaps a little better" than GAIIx data at the end of the read.
The instrument "meets our specifications," Taylor said, and he has recommended its purchase to his institute.
Taylor said he sees several major uses for the platform. Like Hadfield, he believes it will be very suitable to develop assays and to check libraries that will ultimately run on a HiSeq or GAIIx.
[ pagebreak ]
He also thinks the instrument will be used in diagnostics because of its high speed and ease of use. "It can turn data around in a day or two, and it's fairly easy to use, so that will mean that you won't need high-end lab specialists to operate it," he said.
In addition, MiSeq might be useful for the quick development of clinical assays, an area "that probably is not going to be quite diagnostic or quite research, but I think the MiSeq could meet very effectively," he said. For example, it could be used "for those situations where you wanted to develop new assays that are not commercially viable as CE-marked assays … piloting them on a MiSeq as a kind of development, and then those ones that meet the rigors of clinical utility could then perhaps go forward and be made into straightforward CE-marked kits."
Ion PGM vs. MiSeq
MiSeq is the second short-read small-scale sequencer to arrive on the market, following Life Tech's Ion Torrent PGM earlier this year, and their applications are likely going to overlap.
"They do overlap in terms of speed," which is currently fairly similar in terms of yield per day, said Nusbaum.
One advantage for the MiSeq is that it generates a data type that is familiar to existing GA and HiSeq customers, and for which a lot of analysis software already exists.
MiSeq also currently has a "slightly higher" base accuracy, Nusbaum said, but he does not think that difference matters "for a lot of applications."
What might matter more, he said, is that fact that the PGM only provides unpaired 100-base reads at the moment, whereas MiSeq generates paired-end reads of 100 bases and more. That might change, however, if Ion Torrent either develops paired reads or increases its read length, he added.
"Probably, in the end, it's a cost question," Nusbaum said. Because the MiSeq only arrived recently at the Broad, the institute has not yet been able to compare the cost of running the two platforms, but he said they will probably not differ by more than a factor of two.
Acquisition costs, rather than running costs, might be another factor playing into the uptake of the platforms. The list price for the MiSeq is $125,000 while Life Technologies is selling the PGM for $50,000. "It may be a challenge to raise enough money to buy a $100,000 machine, but less of a challenge to buy a $50,000 machine," Nusbaum said.
Hadfield said he and his colleagues considered the MiSeq, Ion Torrent PGM, and 454's GS Junior when they first started to look for a methods development platform. Both the PGM and the MiSeq provided sufficient reads for their needs, but the lab decided to go with MiSeq because it already had experience with Illumina sequencing. "We have the workflow built up in the wet lab, we understand how the sequencing works so we can troubleshoot that very easily, and we have bioinformatics pipelines that are built around the Illumina data," he said.
But Hadfield is still interested in the Ion Torrent platform, especially if its read length increases. "If they can take the technology in the same way that Roche has taken the 454, to 600 to 700 base pair reads, and if they can achieve the read depth of Illumina, the millions or tens of millions of reads from a chip, then Ion Torrent is a very attractive technology," he said, especially for applications like metagenomics that require long reads.
According to Taylor, the quality of the Ion Torrent data he has seen so far is "one or two logs down" from the MiSeq data. "I think that's going to be a problem that needs to be resolved before it's a serious contender," he said.
However, he is watching the development of both platforms closely and could imagine roles for both, for example to confirm each other's results. "Traditionally, one does next-generation sequencing and confirms with Sanger sequencing, but why not do it by Illumina and then confirm with Ion Torrent, or the other way around? I think between them, they could easily replace capillary sequencing."
According to Orpin, Illumina plans to ship "tens" of MiSeq instruments during the current quarter and "hundreds" of instruments in the fourth quarter. Several early users are scheduled to present data and results at the American Society of Human Genetics annual meeting in Montreal next month, and several blogs are expected to post MiSeq results soon.
Have topics you'd like to see covered in In Sequence? Contact the editor at jkarow [at] genomeweb [.] com.