Users of second-generation sequencing platforms from Illumina, 454/Roche, and Life Technologies/Applied Biosystems witnessed improvements in yield, read length, and data quality in 2008, though many struggled with technical problems relating to their instrument or reagents, according to a survey conducted by In Sequence at the end of 2008.
Also, the $10,000 human genome remained elusive in 2008, as the total cost for users to generate a gigabase of high-quality data ranged between $1,600 and $146,500 at the end of the year, depending on the platform.
Almost 60 percent of users are planning to add an additional instrument in 2009, though about 10 percent are putting off additional purchases to wait for the arrival of the third generation of sequencers.
In late November, In Sequence e-mailed six survey questions to approximately 100 users of second-generation sequencing platforms, asking them for feedback about their experience. All participants were promised anonymity.
They were asked about the type of platform they use and the number of runs per year they perform; applications; total cost for a gigabase of high-quality data; technology improvements and systems failures over the last year; and plans to acquire additional second-gen sequencers in 2009.
Twenty-two users responded, representing a total of 76 sequencers from Illumina (47), Roche/454 (18) and Life Technologies/Applied Biosystems (11).
Ten users operate Ilumina Genome Analyzers, 14 users own the Genome Sequencer FLX, and six users run the ABI SOLiD platform. Several users (see table below) own multiple instruments of the same type.
Eight users, or more than a third, said they operate both short-read and long-read platforms, including four who own at least one GA and one GS FLX, three who run SOLiD and FLX instruments, and one who reported operating all three platforms.
Most users of the Illumina and 454 systems have started to run their systems frequently, averaging 52 runs per year for the GA (range: 10 to 100) and 58 runs per year for the GS FLX (range: 30 to 150 per year). Users of the SOLiD system currently run their instruments less often — 28 times per year on average (range: 15 to 50+), possibly due to a longer run time, or because they did not run the instrument as frequently after setup.
In order to gauge what the application "sweet spot" for each instrument type might be, In Sequence asked users what applications they run on their sequencers. As it turned out, all three systems are being used for a variety of applications, indicating that at this point, they are not confined to specific niches. However, the GS FLX appears to be used more heavily for de novo genome sequencing than the other two systems at the moment.
For the Illumina GA, the most-cited application was genome resequencing, followed by chromatin immunoprecipitation sequencing (ChIP-Seq) and mRNA/cDNA sequencing. Users also mentioned digital gene expression, de novo genome sequencing, amplicon sequencing, methylome sequencing, synthetic DNA sequencing, and DNAse hypersensitive site sequencing.
For the GS FLX, which offers longer reads than the GA or the SOLiD, the most cited applications were de novo genome sequencing and mRNA-Seq/cDNA sequencing, followed by genome sequencing (not specified whether de novo or resequencing), amplicon sequencing, and genome resequencing. Participants also mentioned metagenomic sequencing and BAC sequencing.
SOLiD users most often listed genome resequencing and small RNA sequencing, followed by ChIP-Seq, transcriptome sequencing (they did not distinguish between mRNA Seq and digital gene expression), and de novo sequencing.
Total Cost of Sequencing
Next, users were asked to estimate the total cost for generating a gigabase of high-quality data. They were asked to include — and break down, if possible — all costs, such as labor, sample prep, sequencing consumables, instrument amortization, service contracts, data analysis, and data storage.
Not surprisingly — since instrument and reagent prices are negotiable, and labs differ in how they calculate certain costs — the results varied widely for each platform, although differences between the three became apparent.
Based on results from seven respondents, a gigabase of quality data on Illumina's platform costs $6,700 on average (range: $1,350 to $12,000).
All three users who broke down the contributing factors listed consumables — for sample prep and sequencing — as the largest cost that makes up about half of the total, followed by instrument amortization as the second-largest factor that ranged between 30 and 40 percent of the total.
On the 454 platform, the total cost of generating a gigabase of data was considerably higher than for the GA, and differed markedly between labs using the "old" FLX chemistry and the new Titanium chemistry, which 454 launched last fall.
Five 454 owners who have been using the Titanium platform reported an average total cost of $24,400 (range: $17,500 to $29,000) for a gigabase of quality data.
Four customers who have been using the "old" FLX chemistry, on the other hand, reported an average total cost of $102,200 per gigabase of data, with an upper end of $146,500 (the user who reported the lowest cost asked that this number not be revealed).
Reagent costs made up the biggest chunk, according to three 454 users who broke down the total cost, ranging from 62 percent to 80 percent.
A gigabase of quality data generated on the SOLiD system was the least expensive based on information from three users who responded and reported an average total cost of $1,700 (range: $1,600 to $2,000). However, one of these users said the quality of his SOLiD data was "very shaky." Only one user broke down the cost, attributing 90 percent to consumables for sample prep and sequencing.
Next, we asked users how much their platform has increased in read length, yield, accuracy, and run time over the last year.
For the Genome Analyzer, nine users, or 90 percent, reported improvements, which some of them attributed directly to the switch from the GAI to the GAII and new reagent kits released over the year.
Specifically, two users reported improvements in accuracy, which increased from 98.5 percent to more than 99 percent. Six users said the yield per run went up, in one case up to 12 gigabases. Five users noted that the run time shortened by between 25 percent and 50 percent. Three users reported improvements in read length — to 36 base pairs, 40 base pairs, and 88 base pairs, respectively.
While two users noted that long reads are "very important" to them, two others said they prefer paired-end capabilities or more reads over longer reads.
For the GS FLX, five users, or 36 percent, said that the new Titanium chemistry had improved their yield per run, with read length increasing to 400 base pairs from 250 base pairs. Three users said that they did not yet have access to the Titanium upgrade, but were expecting to use it in 2009, and two users said they had just started initial runs with Titanium consumables.
Three SOLiD users, or 50 percent, said they had witnessed significant improvements in yield, read length, and data quality over the year, coinciding with the switch from the first version of the instrument to SOLiD 2.0. Two users said that the yield per run increased 10-fold. The read quality, especially at the 3' end, "has increased substantially" over the last nine months, according to one user, allowing read length to grow to 50 base pairs from 25 base pairs.
Besides improvements, we asked users how often and for how long their system had failed over the last year, and what the causes of these failures were.
Nine Illumina GA users, or 90 percent, said that their systems had been down over the last year. While two users said their system failed just once, four users reported that their instruments were down between 10 and 20 percent of the time, and the remaining three cited several failures.
Six users cited instrument-related problems as the root of their trouble, five complained about reagent quality issues, three mentioned human error as the reason for failure, two said service work required downtime, and one blamed the lack of availability of a cluster kit.
Seven GS FLX users, or 50 percent, said they experienced failure or downtime over the last year. Two of them reported a single instrument failure, while the others cited more frequent problems. Two users said they had no failures or downtime at all.
Four users said hardware or equipment failure caused the problem, three pointed to reagent quality issues, and one cited human error related to the production of high-quality sequencing libraries.
Among SOLiD users, two thirds, or four, reported problems, all of them more than one. Two users cited hardware or mechanical problems, one pointed to reagent quality problems, and one said software errors contributed. One user said he had not had any problems since the instrument was installed two months ago.
Planned Purchases in 2009
Finally, we asked users whether they are planning to acquire additional second-generation sequencers in 2009, and which systems — including those that are not yet commercialized — they are considering in their planning.
Thirteen users, or almost 60 percent, said they will purchase another system this year, while six said they were not planning to, and three were still unsure.
Five users said they are considering to purchase a Genome Analyzer, among them two that already own such an instrument; four are considering SOLiD, among them two current SOLiD users; and two are thinking about a GS FLX, both of them current 454 users. One lab is considering to obtain a Polonator, and one lab is hoping to become an early-access user of Pacific Biosciences' system this year.
Four users stated that they are not yet including the third generation of instruments in their planning. "It is too early to tell whether they will offer significant improvements over our current equipment, given the current rate of improvements in generation two," according to one user. Another one stated that he looks forward to the third-generation platforms, "but I don't waste a lot of time planning on systems that are still 'vaporware.'"
Two others, on the other hand, said that they are not going to acquire any new instruments this year because they are waiting for the next generation of sequencers to arrive, one of them mentioning Pacific Biosciences in particular.