I am not attending the American Society of Human Genetics meeting in San Fransisco, but can’t escape the buzz it creates on twitter (hashtag #ashg2012). Strikingly, it is almost another AGBT when it comes to announcements from companies selling sequencing instrument. All of them had something new to bring to the floor. This post summarizes what I picked up from twitter and a few websites, and I give a bit of my perspectives on the respective announcements. I am focussing on technology improvements, especially with regard to read lengths, not so much on applications such as cancer resequencing panels.
After their splash announcement at AGBT, it was very quiet for 10 months. Now, at ASHG, attendants get to see, and touch, the MinION and GridION. Pictures were all over twitter, collected here. @erlichya posted an annotated picture of the minIon on twitter.
However, still no data has been released for us researchers to judge the platform by. The Chief Executive Officer Gordon Sanghera is quoted in the Nature News blog post as saying “What we said at AGBT, we will make good” – but without any timeline. The proof is in the pudding, and the lack of it frustrates me and many others. For instacne, have a look at this excellent post by @mikethemadbiol.
Roche/454 Life Science
Their headline was “Roche introduces new GS FLX+ software for significantly improved long-read sequencing performance”. This is a software upgrade, and a bit of tweaking with the instrument, to finally get GS FLX+ to work. It has taken them over a year after the introduction of the GS FLX+, and many labs – including ours – were frustrated by the lack of consistent performance (see my rant at the beginning of this year). We have been able to try a few runs with this upgrade, and so far it seems to be working. The E. coli test run (library and beads prepared by 454 Life Sciences) showed an incredible peak (mode) read length at 950 bp. A mode of 800 bp could now be the norm. No change in the number of usable reads, though. Next up is making the long reads working for amplicons.
According to twitter, the market leader announced 2 x 300 bp read lengths for the MiSeq, with and even longer reads coming MiSeq in 2013. @BioMickWatson added on twitter “they announced (2x) 400bp at Illumina users meeting” and even longer reads in R&D. If this materializes, around 750 bp amplicons could feasibly be sequenced on the MiSeq. No word, though, on longer reads for the HiSeq (or the rapid run mode on the HiSeq 2500).
Life Technologies was – as usual – boasting about what is to come. Superlatives were flying, such as Jonathan Rothberg’s statement that they are “bending the laws of physics”. What really caught my attention, though, was, again from Rothberg, “PGM now has the industry’s most accurate technology for homopolymers, now at 99 percent accuracy.” 99% accuracy, wow. Can they back this up with data? What does this actually mean? For what length of homopolymer? Is this per-read, consensus? So, more questions than answers here. Benchmarking, anyone?
The first user reports on the Ion Proton indicated routinely 9 GB, and up to 10.8 Gb throughput (at 200 bp) per run. Ion/Life are usually very forthcoming with sharing datasets – albeit behind a (free) login through the Ion Community. Surprisingly, there still is no Proton dataset available (do we as a community need to worry why this is?).
Finally, Avalanche, the hands-off, emPCR free sample prep instrument, was announced (again), but it remains to be seen when that will materialize (apparently in 2013). It must be said, though, that it looks like Ion/Life is delivering more or less what they promise when they promised it.
PacBio timed the release of their ‘XL product enhancement‘ with ASHG. New chemistry and software have the average read length go from 3500 to 5000 bp, with a faster polymerase. Maximum read lengths increase to 20 kb. Also, the maximum movie lengths is increased from 90 to 120 minutes. No increase in per-base quality, though.
Ion Torrent and Illumina
These two platforms are in an eternal battle – they nicely had their workshops in parallel, resulting in a kind of real-time ‘tweet war‘ not seen before. Illumina is not lagging behind when it comes to read lengths, in fact, they have the advantage: currently, 2×250 (paired end) on the MiSeq wins over 300 bp single end on the PGM. Even the coming 400 bp reads on the Ion won’t make a difference here. No paired end may be a killer (although Ion released a protocol requiring the chip to be taken off the instrument and placed back on for the second read). The MiSeq is already emPCR free, and very easy to use. Still, Life Technologies seems to be selling more PGMs than Illumina is getting MiSeqs out of the door. It is a neck-and-neck race – both companies forced to increase throughput (numbers of reads as well as read lengths) to not lose out on the other. Let’s hope they keep quality as their main driving force!
PacBio and Oxford Nanopore
Both these are single molecule sequencing platforms, but with one difference: PacBio has a working machine sold to a (not too big) base of customers. Oxford Nanopore hasn’t even gotten machines out to early access sites, and the whole twitterverse is crying out for data to look at. PacBio is working hard to increase their read length, this is giving them a head start in this market – although the high instrument price and low per-base quality is making community-wide acceptance difficult. If, however, Oxford Nanopore delivers what it promises, it will win at the very least on throughput and price per base, if not on read length.
Roche/454 Life Science and the rest
What niche for 454 GS FLX+? Yes, the longest high-quality reads available, but there still is no substantial movement when it comes to the price per base. The platform has just too expensive run costs to become much bigger than it is now. It can’t beat PacBio on read length – although the instrument is cheaper. It is way (way!) behind Ion/MiSeq on throughput and price per base – although it does produces very nice Sanger-like reads. Roche/454 suffered a lot from the inconsistent GS FLX+ results in the hands of many users, and it remains to be seen if they make up for this with the now, hopefully stable running, long read instrument.
Did I miss anything? Do you disagree? Let me know in the comments section?