What's in a PacBio Revio?
The original Pacific Biosystems RS was a big, complicated instrument which likely had a high cost-of-goods (COGS). From surplus PacBio parts I’ve seen on eBay, it likely used a couple of Andor Neo sCMOS cameras. I would guess at the time these were somewhere in the $100000 range.
Later instruments (Sequel) are likely have a much cheaper COGS. By moving much of the optical system and sensing to a consumable chip they vastly simplified the instrument.
The new Revio weighs 449Kg about 100Kg more than the Sequel II. With a relatively simple optical system, I’m curious to know exactly what’s going on in these instruments and how high the mark up is on that $779000 price tag. If you have any further information, please get in touch (new@sgenomics.org, or over on twitter).
The Revio likely has at least the following major components:
Fluid handling system (pipette robot)
Optics, including lasers and alignment camera
Anti-vibration platform
Data acqusition hardware
Primary data analysis compute
New fancy Nvidia GPU compute for machine learning
Cooling System
Racks and chassis
This is in part supported by internal images of the Sequel:
Weight
I thought it would be interesting to go though the above components and see how much of the overall weight we can account for:
This more or less would account of the weight of the Sequel, but leaves ~100Kg from the Revio. Given that a lot of the internal frame on the Sequel II is also metal this could account for much of this. But please get it touch if you have any thoughts!
Optical System and Cost Estimates
The Revio presentation shows a reasonably close up image of the chip based consumable:
This looks like an evolution of the Sequel II chip (no doubt a huge amount of development work went into increasing sensor density, but I’d guess the basic approach is similar.
And it looks to be a reasonable match for recent patents, which show chips with a single optical inlet:
And I assume these use the same chip embedded fresnel lens structure and filters we’ve previously seen. The patent also shows the overall optical system:
The patent suggests the use of a 520nm diode laser with ~2nm FWHM and ~61mW power. In small quantities similar diodes cost $76. More broadly 500mW DPSS 532nm for scientific applications are in the $10000 range, in the estimates below I doubled this to account for other optical components.
My full cost estimates are behind the paywall break below. But my rough “guesstimate” based on public pricing is $100,000.