After undergoing a series of internal feasibility studies, Pacific Biosciences has determined a number of specifications for the first version of its single-molecule real-time sequencer, which it plans to start shipping to customers in the second half of 2010, according to a company official.
During a presentation at the Personal Genomes conference at Cold Spring Harbor Laboratory last week, chief technology officer Steve Turner also revealed how the company has further developed and tested its technology, including disclosing for the first time details about two sample-prep methods for its system, and results from an internal sequencing project on bacteriophage φX174.
A noteworthy feature of the system, desired by many potential customers, will be the ability to give users a choice between different read lengths and throughput levels, according to CEO Hugh Martin, who spoke with In Sequence last week.
Both parameters are interdependent, since at the beginning of a run, a greater percentage of polymerases is active and generating data than later on in the run. For example, in a resequencing study, a user might want to increase the throughput by dialing down the read length, while for a de novo sequencing project, the user “can essentially turn a knob, and the system will deliver much longer read lengths” at a lower overall throughput, Martin said.
“Today, the way customers solve that problem is, they buy a 454 and an Illumina,” he said, which requires them to operate two hardware systems and use different protocols. By comparison, PacBio’s platform, called the SMRT Sequencer, will be able to perform both tasks.
The maximum read length of the initial system will be “equivalent to, or greater than, what you would get with the Sanger system,” or between 800 and 1,000 base pairs, Martin said.
He said that PacBio has not yet determined how many reads per experiment its instrument will generate, or what the throughput for a given read length will be. That will depend, for example, on the performance of the chemistry, the number of holes on the chip, and their occupancy with polymerase, said Martin.
The zero-mode waveguide chip used by the current development-stage instrument has 3,000 holes, but that installed on the commercial system “will be significantly bigger,” said Martin, adding that faster cameras and stronger illumination will be required to record data from a larger chip area.
Also, while only about one-third of the holes are currently occupied by a single polymerase enzyme — and are therefore useful for recording a single-molecule sequencing reaction — the company has several technologies under development that will increase that rate, he said.
Users “can essentially turn a knob, and the system will deliver much longer read lengths.”
The commercial system will be able to sequence genomic DNA, cDNA, or PCR products and accept linear or circular template DNA, said Martin. The polymerase will incorporate at least three nucleotides per second. Since the reactions are measured in real time, the speed of the system will be high: at short or medium read lengths, an experiment can be completed within approximately five minutes, according to Martin. However, users will be able to load batches of samples, so the system can run unattended for up to eight hours.
The consumables cost of such a five-minute experiment is expected to be several hundred dollars, he said. The price of PacBio’s instrument will be “in the range of the other next-gen players,” with the exception of Helicos BioSciences and Danaher Motion. That would put its price between approximately $430,000 and $600,000, which is the price range of the Illumina Genome Analyzer, 454 Life Sciences’ Genome Sequencer FLX, and Applied Biosystems’ SOLiD system.
Also last week, PacBio for the first time revealed details about sample-prep methods it has been developing for its system.
The first method, used internally for “quick projects,” enables researchers to obtain sequence data in less than an hour. Driving this are random primers that are hybridized to single-stranded circular template DNA at certain intervals, which the polymerase uses to start synthesizing DNA.
The company has used this method, for example, to sequence φX174, a bacteriophage with a single-stranded circular 5.4-kilobase genome, Turner said during the Cold Spring Harbor conference last week. Though the genome sequence produced using this sample prep method had no gaps, the coverage was not uniform, he noted, probably owing to the nature of the random primers.
Turner also described a new sample-prep method that takes between four and six hours and allows the company to turn long double-stranded linear DNA into a single-stranded circular template. To achieve this, the scientists ligate one single-stranded hairpin adaptor to each of the two ends of a double-stranded DNA fragment, turning it into a “dumbbell” structure, which PacBio calls “SMRTbell.”
DNA SMRTbells are structurally linear, but topologically circular, Turner pointed out. A polymerase enzyme can start sequencing from a primer hybridized to the single-stranded adapter, opening the double-stranded DNA as it synthesizes the new strand.
According to Turner, an advantage of this approach is its ability to obtain sequence information from both the sense and the antisense strand of the original double-stranded DNA in the same run. Also, because the template is circular, the polymerase can go around several times, thereby increasing the consensus accuracy.
PacBio has also found that the method, which the firm has patented, increases the loading efficiency of large DNA circles into zero-mode waveguides “because it’s effectively a stick,” according to Martin.
Using the SMRTbell sample-prep method, company researchers have sequenced the φX174 genome with more even coverage, Turner reported. The sequence differed in four positions from the reference genome: three of these turned out to be variants in the strain sequenced, while the fourth discrepancy is still under investigation.
Excluding this error, the scientists found that errors in the consensus sequence decreased exponentially with the depth of coverage, according to Turner. The sequence was error-free starting at 13-fold coverage, and to within the resolution afforded by this small genome, he and his colleagues found no systematic errors.
At the conference, Turner also addressed how PacBio has been improving the system’s raw accuracy. Insertion errors occur when the polymerase binds to the correctly labeled nucleotide but releases it without adding it. However, the system erroneously records the event as a base addition.
The company has now reduced this error rate by combining amino acid changes from two different mutant polymerases that incorporate correct nucleotides more frequently without releasing them.
Deletion errors, on the other hand, happen when the polymerase incorporates a nucleotide too quickly for the camera to record the event. That problem might be remedied by evolving the enzyme to hold the nucleotide in the active site for a longer period of time, thereby insuring that the camera is able to scan the signal.
Following the bacteriophage project, PacBio has already embarked on other sequencing projects, and plans to present new results at the Advances in Genome Biology and Technology meeting in February 2009, according to Martin.