Tuesday, October 27, 2015

BGI Launches the BGISEQ-500

This weekend brought the formal launch of the BGISEQ-500 desktop sequencing instrument from BGI (though deliveries won't begin until early next year).  Utilizing the ex-Complete Genomics ligation technology also used in the Revolocity system, the instrument appears to sport a price similar to the Illumina NextSeq but offers throughput somewhere running from a NextSeq up to the low end of a HiSeq. Two flowcells can be run at a time (apparently in sync with each other, unlike QIAGEN's long-delayed machine), with a small and large versions of the flowcells.  There's some ambiguity on questions such as the precise read length, though it is very short compared the the typical Illumina offerings.  Dale Yuzuki has a nice write-up (complete with a picture next to the box) based on attending the International Congress of Genomics 100 where it was unveiled.  One of these days I should wangle my way to that conference -- not only would the genomics be fascinating, but China holds a special allure -- or more specifically, Ailuropoda, for our household.

BGI is also offering a library prep instrument which can perform a wide array of assays, though the details on this are even sketchier.  What caught my eye is the Long Fragment Read (LFR) protocol, which was developed by Complete several years ago and could potentially compete with scaffolding preps such as 10X and Dovetail.  LFR is a great example of an interesting technology that would probably be better off spun-out, being locked up with Complete Genomics/BGI has meant LFR has never captured much use, despite being far earlier in the field than competitors such as Dovetail, 10X and Moleculo.

As an aside, I think the BGISEQ-500 is a splendid example of why standardized calibrator DNA samples are needed along the lines of my last post. Given that the cPAL technology involves ligating sets of probes to the target, it would be very valuable to understand how well this approach performs on homopolymers.  Even of perhaps greater interest is how well a short read technology, combined with a range of paired end libraries, can perform in assessing Simple Tandem Repeats (STRs) that are significantly longer than the read length. My guess is that this is a complete mismatch of technology and need, but proving that out would be valuable.  Parenthetically, if anyone wants to discuss the calibrator design concept privately, my email address is keith.e.robison over at Glycine-mail (or is it Guanine-mail?  I can never remember).

I really should try to learn more about the new instrument (though the short reads are a severe handicap for my dominant usage of de novo assembly), but there's an obvious but quirky excuse not to: BGI has announced that BGISEQ-500 will only be sold in China. That doesn't entirely rule out testing the device - Chinese companies are serious players in the sequencing outsourcing market, if you can accept the inherent delay in shipping samples across the Pacific (and learning of a number of specifically Chinese holidays).  Unless things have changed, another quirk is a prohibition on shipping dry ice packages to the mainland, though Hong Kong will accept them (which I suspect was a component in BGI's decision to locate their outsource sequencing center there).

BGI hasn't said much for the reasons for this national exclusivity.  I can think of a number of possible reasonss, which aren't mutually exclusive and could form part of the reason.

One possible reason would be patents; even if BGI thinks they have good IP coverage the possibility of doing battle in U.S. or European courts might simply be unpalatable.  Hardly the first time this has happened; Nimblegen located their array processing facility in Iceland precisely to take advantage of a lack of patent coverage for Affymetrix there (though I can't remember seeing an explanation how Iceland ended up uncovered).  However, this seems unlikely given that most of the same technology is in Revolocity, which recently placed several more units around the world.

Launching only in China could be simply a conservative strategy, given the challenges in launching a new system and supporting it with reagents and training. In this case, perhaps 6 or 12 months down the road a wider release would be in the cards.  It could also be a point of national pride, given that the instrument is described as being developed and manufactured in China. 

In any case, 2016 is already looking like a bumper year for new and revised sequencing instruments, with the BGISEQ-500 joining PacBio's Sequel and Oxford Nanopore's PromethION, with perhaps QIAGEN's sequencer (based on an acquisition made 3 years ago!) and maybe even Roche/Genia or BIO-RAD/GnuBio joining them?  There's even a new story (alas, only on GenomeWeb Premium), of another Chinese company resurrecting Helicos' technology.  How hot is genomics instrumentation?  So hot that former zombies walk the earth again!  If that's still true after Halloween, then 2016 should be a very interesting year for those of us who gawk at genomics and squawk about it.

1 comment:

Dale Yuzuki said...

Hi Keith, nice to hear your perspective on BGI's new benchtop NGS system.

On the readlength, a few clarifications that came up while chatting with Radoje Drmnanac. (Believe it or not, after stepping out of a plenary for a prospective partner meeting then hanging out with Suzanne Yokota at the Complete Genomics booth not only was I able to meet Rade but also Maynard Olson.) The Revolocity is 2x28 bp mate-pair, with 375 ave insert length but could go up to 1kb.

The BGISEQ-500 as announced is offered as 1x50, 2x50, 1x100 and 2x100 bp readlengths. They have also changed the name of their chemistry from cPAL to cPAS - combinatorial Probe Anchor Synthesis, and a slide (that I wasn't able to capture nor got a translation in the audio as Xu Xun gave the entire presentation in Chinese) made it clear it was polymerase-based, not ligase-based.

I've been told since writing up my original post (and should update the original one) is that the system will be priced 1/3 less than 'an equivalent sequencer' which is the NextSeq-500. or about $165K, and the cost of the reagents also 1/3 less.

Whereever this technology goes will be interesting, and something to keep a close eye on for sure.