It is important to note the performance spec - it's a bit less in total data than Revio and that is deliberate. Partly it is to keep the Revio attractive, but it also represents a compromise for keeping the instrument price down - the compute on the Vega is sized to process 60 gigabases of data in the specified time and running more data would require proportionally more compute - and these days the fancy GPUs or FPGAs are a major cost driver for the instruments. In my conversation with CEO Christian Henry, he pointed out that effectively Vega is skimming the best data, so 60 gigabases of Vega data isn't quite equivalent to 90 gigabases of Revio data. PacBio plans to ship Vega in earnest first quarter of next year and is hoping to place multiple instruments late this year - if you're at the party (I'm not) you can place your order there.
On the workflow side, Vega is fully compatible with all PacBio library prep products; there isn't a chemistry change. Note that is this is the older chemistry, and not the SPRQ chemistry that was just announced and is covered in my morning piece. Trying to fit a new chemistry (and associated software) onto a new instrument is taking on a great deal of development risk, and PacBio chose not to do that. Henry wouldn't commit to when SPRQ would move onto Vega, but suggested it probably would happen. Vega maintains the trend of performing the major early analyses onboard - HiFi generation, DeepConsensus with 5mC detection and demux are all performed on the instrument. On test datasets, Vega data was effectively indistinguishable from Revio data.
It's not just for full human genomes either. With PacBio's PureTarget enrichment kit for short read refractory regions of the human genome, 48 samples could be loaded into a single Vega run.
While Henry said upfront "it isn't a MinION" in terms of transportability, for a board demo it was wheeled into the boardroom and set to sequencing. The instrument is easy to load, with lights guiding the user to install two consumable components into a tray which slides out from the center. Vega can also be loaded with a new set of consumables and library mix halfway through a run, so loading a library on Friday to run into Saturday is an option.
Henry noted that PacBio has still been selling Sequel IIe and does not have an end-of-life plan yet for that product - but clearly it would be a strange decision to purchase one now that Vega is offering superior performance at such an attractive price.
It's amazing how the affordability of sequencing constantly improves. MiSeq launched a bit over a decade ago for just a slightly lower price. There's many reasons (such as counting applications) where this isn't a very useful comparison, but the new MiSeq i100 delivers half the data for more than half the purchase price. Or to put in my favorite metrics, when at Warp we started contracting to use the original PacBio RS I for actinomycete sequencing, we required 6 flowcells per bacterium to generate 50+X coverage of noisy Continuous Long Read (aka subread) data; if I could time travel a Vega back to those days it could generate perhaps 400 actinomycete genomes per flowcell -- that's serious productivity gains!! And with the size of it, even in our original tight space I might have been able to make a play to get one.
I believe the logic of Vega will appeal to many core labs and small biotechs. Revio is a ravenous monster and even with the new dropped pricetag, a lot of grant money or seed capital. If you can't feed the beast, why pay to have one? At a quarter-ish the capital cost (or go for the reagent rental model to further reduce upfront cash), Vega is still a very capable instrument for a core lab that sees demand for a few genomes per week or similar-sized batches of microbes, Kinnex, PureTarget or other library types. That also means a lab with significant sequencing demands and an impatient sequencing chief - such as a Warp Drive 2.0 - could have an instrument without tying up large amounts of space or capital. So someone developing a new diagnostic kit requiring HiFi data quality or a plant trait improvement startup or some other sort of biotech bootup company could more likely swing a Vega than a Revio and so gain speed and process control that just isn't available with a good outsource vendor.
I missed ASHG (if the choosing last June had gone badly, I definitely would have been there). Hoping some of the vendors reserve some exciting announcements for AGBT, as I just was confirmed for attending that confab.
No comments:
Post a Comment