This article was originally published Nov. 9.
SAN FRANCISCO — Following the launch of its Ion Proton desktop sequencer in September, Life Technologies officials said last week that more than 7,000 runs have been completed on the system in the company's own labs and in the labs of its customers.
One such customer, Joe Boland of the National Cancer Institute, reported during a workshop at the American Society of Human Genetics conference here this week that his lab has done 50 runs on its four Proton machines and is routinely generating between 8 gigabases and 9 gigabases of data per run with read lengths of around 150 base pairs.
He said that when the machines were first installed in September, throughput was 4 to 6 gigabases and reads were 100 bases, but this rapidly increased to 8 gigabases of throughput and reads of 150 base pairs. Already, the team has had some runs over 10 gigabases and expects to routinely generate between 9 gigabases and 11 gigabases by the end of the year.
Accuracy, too, has improved, Boland said, and is around 99 percent, which is "absolutely fine for what we need."
The NCI lab also houses two Illumina HiSeq 2000 instruments, one MiSeq, and six Ion Torrent PGM machines.
Boland said he planned to use the Protons primarily for rapid exome sequencing, as well as "high-level" amplicon sequencing of between 500 and 1,000 samples, and whole transcriptome sequencing.
The bulk of NCI's sequencing is exomes, Boland said, and the lab's two HiSeq instruments produce 56 exomes per machine every six to eight days, so "our pipeline is constantly full."
The Protons will be used to both increase capacity as well as to sequence exomes for which a rapid turnaround is needed. Additionally, if there are technical difficulties with a sample on the HiSeq it can go directly to the Proton, instead of going back into the HiSeq queue.
Boland said that the center's first exome runs on the Proton using Nimblegen capture array covered about 80 percent of the exome at 20-fold coverage. As the lab continues to gain experience, he said, the percent of the exome covered is continuing to improve, and the lab is now covering just below 90 percent of the exome at 20-fold coverage. Average coverage is around 100-fold.
The team has also tested transcriptome sequencing on the Proton. Using version 2 of the Ion Total RNA Kit and between five and 500 nanograms of input poly-A RNA, he was able to generate 50 million reads from a melanoma cancer sample, compared to just 2.5 million reads using the PGM 316 chip.
Going forward, Boland said that he is looking forward to sample-prep improvements and automation with the Avalanche sample prep chemistry and the Ion Chef system, products that Life Tech is planning to launch in 2013 (IS 9/18/2012).
Avalanche is Ion's emulsion-free PCR sample prep chemistry, which it plans to launch next year. Similar to the company's Wildfire technology for its SOLiD systems, Avalanche will based on isothermal amplification. Meantime, Ion Chef is a system for automating sample prep, reducing hands-on time, and thus increasing the efficiency of chip loading.
"That consistency will be key," Boland said of the Ion Chef.
Additionally, Boland said that the lab plans to expand into whole-genome sequencing on the Proton by mid-2013.
Other users are having similar experiences with their Proton machines. Richard Allcock, an associate professor at the School of Pathology and Laboratory Medicine at the University of Western Australia in Perth said his laboratory received its Proton in October.
He said that his lab is currently generating about 7 gigabases of data per run with read lengths of 100 to 110 base pairs. The lab has only done about five runs on the system using control libraries generated by Life and two runs using its own libraries.
The laboratory also has two PGM machines and he said that the data quality is comparable between the two instruments, although the Proton is "more sensitive to workflow issues" than the PGM,and loading efficiency is currently around 80 percent on the PI chip.
He said that he considers a system to perform accurately if at least 60 percent of the sequence has a Q-score of 20, and so far all the Proton runs have been at that level.
Moving forward, he said he would like to see improvements to the read lengths. He said that he had expected reads on the Proton to start at the level of read lengths on the PGM, "so to hear that read length started at 100 base pairs, that imposes a lot of restrictions."
Aside from the two PGMs and one Proton, the lab also operates two SOLiD 5500 machines and previously ran a 454 GS FLX machine. Over the next year, he said, the lab will likely phase out its 5500 machines and run the two PGMs and one Proton machine.
Allcock said he is aiming to run around 200 exomes on the Proton in 2013, and also plans to convert a 300-gene neuromuscular panel that is currently running on the SOLiD to the Proton.
The combination of the Proton and PGMs will offer "massive flexibility" to sequence everything from small genomes and targeted panels, to exomes and large panels, and eventually whole genomes.