This article has been updated from a previous version to include updated specs from PacBio on the average read length for its sequencer.
By Monica Heger
Nearly a year after Roche's 454 Life Sciences launched the GS FLX+, which promised modal read lengths of 700 base pairs with the longest reads reaching more than 1,000 base pairs, a number of customers are still experiencing problems with the system.
"On paper it's nice, but it takes a long time to get to work," said Lex Nederbragt, who coordinates sequencing on the 454 and Pacific Biosciences systems at the Norwegian Sequencing Center in Oslo. Since purchasing the upgrade in August, he said his lab has consistently been having problems getting good runs, although he said they are slowly improving.
For the first four months of operating the GS FLX+, Nederbragt said the instrument was averaging reads of half the length of those obtained with the Titanium chemistry and half the throughput.
"It's really frustrating," he told In Sequence, "because we were hoping to ramp up the use of the 454 with long reads."
After a service visit in January, the lab has been "slowly getting longer reads." It recently started to achieve modal read lengths of more than 700 base pairs, but the throughput is still low, he said.
Nederbragt said that the lab previously generated "well over 1 million reads" from one run on the GS FLX, but with the upgrade, it is struggling to reach 1 million reads.
He said the problem appears to be a hardware or instrument problem, and "after the last service visit we are seeing improved results, but we still seem to be behind in throughput."
Stephan Schuster, a professor of biochemistry and molecular biology at Penn State University, said he received the FLX+ instrument last October, and "like many other people, we did not have instant success."
At the time, the lab was working on a large plant genome and wanted to use 454's long reads to help improve the quality of the genome and assembly.
Because the FLX+ upgrade was not working in the lab, the team ended up using 454's sequencing service center to generate long read data to complement sequencing data from an Illumina system and the 454 GS FLX with Titanium chemistry. Schuster said the 454-generated long read data was of very high quality and improved the assembly immensely.
"When we doubled the data using the FLX+, the assembly statistics completely changed," Schuster said. "It was clear we were now assembling a part of the genome that we did not target with all of our Illumina sequencing and Titanium sequencing that we had before. With the FLX+, you are much more completely covering the genome."
However, he said, "we had hoped that we would generate the data ourselves."
Schuster said he has been a 454 customer for years and "this clearly is the most difficult transition they have made to a new sequencing chemistry."
He said that the problems seem to be due to multiple factors, including the instrument performance itself, the sequencing chemistry, and the software.
Because Schuster has been working with 454 to troubleshoot the problems with the platform, he is under a confidentiality agreement and could not elaborate further on the specific issues with the system.
However, he said that the lab exchanged one of its instruments, which has helped improve the sequencing runs, and "just over the last few weeks, with really intense attention from 454, we have made progress, and reached all the specifications that they say is a pass for the sequencing."
Nederbragt and Schuster's don't appear to be isolated cases. Asked about their experience with the GS FLX+, several sequencing service providers, including Microsynth in Switzerland and LCG Genomics, IMGM Laboratories, and Eurofins MWG Operon in Germany told In Sequence this week that they had experienced problems with the upgrade, which in some cases have not been completely resolved yet.
Two service providers said that in initial test runs, the GS FLX+ performed according to 454's specifications but that in subsequent runs, read lengths declined.
A spokesperson for 454 said in an e-mail that the company has "received feedback from some customers regarding the performance variability of the GS FLX+ System post-upgrade," and that it is its "highest priority to resolve those issues rapidly."
The company has since identified several areas of improvement, which will include "minor hardware changes, as well as a software upgrade," and these developments should "improve the overall performance of the system, particularly the read length."
The spokesperson added that at least some of the issues seem to have come from increasing the number of flow cycles. The GS FLX+ instrument required increasing the number of flow cycles in a sequencing run from 200 to 400, which has caused a "greater level of absolute run-to-run performance variability."
Instrument hardware improvements and software improvements should address the variability, the spokesperson added, though the company did not provide a specific timeline for the release of these improvements.
Schuster said that despite these problems, the long reads are still a promising technology and could be a boon to the company as it faces increasing competition from other sequencing vendors, such as Pacific Biosciences, which touts long reads — an average of more than 3,000 base pairs with some nearly 15,000 base pairs with its C2 chemistry — as a key selling point of the PacBio RS. Even Illumina and Life Technologies are steadily increasing the read lengths of their short-read sequencing platforms.
454 "could get a second life here, if they get [the long read chemistry] to work," Schuster said. He said the quality of 454 sequencing data has been steadily improving and despite the increasing read lengths from the Illumina HiSeq and MiSeq platforms, and the projected 400 base pair reads promised by Ion Torrent's PGM, there will still be a place for 454 in the market.
Additionally, he added, 454 has the potential to "massively profit" from PacBio's so far limited success. As of February, PacBio had installed just 48 systems (IS 2/14/2012).
Furthermore, Schuster said that he is aware of unpublished work comparing the two instruments. Using the latest chemistries on both instruments, these groups have found that to achieve equivalent results of the same quality when resequencing small reference genomes, much more data has to be generated on the PacBio, making it more expensive, he said.
Nonetheless, some customers have opted for the PacBio technology over 454 for long-read applications. For example, a lab at the Department of Energy's Joint Genome Institute is currently in the midst of phasing out its 454 instruments in favor of PacBio (IS 11/8/2011)
Sequencing plant genomes could be a key application for the GS FLX+ instrument, Schuster added. Plant genomes tend to be large, polyploid, and highly repetitive, making the long reads essential for covering the entire genome.
"In our lab, the perfect combination for us is a mixture of HiSeq, MiSeq, and 454," Schuster said. He said the long reads and the two different chemistries can generate genomes that are of even higher quality than those generated from Sanger sequencing.
However, he added that it will be critical for 454 to address the problems customers have been having with the FLX+, especially because other platforms like Ion Torrent have made rapid progress with increasing read lengths.
"It's clear that if the FLX+ is not robust and stable, and stays at 400 base pairs, then Ion Torrent will catch up to them," he said.
Nederbragt agreed that 454 will face increasing competition from Ion Torrent and the Illumina MiSeq as both companies continue to make improvements to read length on their respective instruments.
Nederbragt said his lab uses the 454 machine for amplicon sequencing applications that require reads longer than 400 base pairs, but noted that as read lengths increase for the Illumina MiSeq and Ion Torrent PGM, more of the market will migrate to those platforms.
While "it's still a bit early" for the Illumina and Ion Torrent platforms to replace the 454, Nederbragt noted that "in terms of price per experiment, 454 is definitely being out-competed."
Nevertheless, he added, "I'm not willing to trade in our 454 yet."
Have topics you'd like to see covered by In Sequence? Contact the editor at mheger [at] genomeweb [.] com.