Sunday, December 18, 2016

Sequencer Archaeology

One of the frustrating side effects of outsourcing all your sequencing is that lack of connection to the actual machines.   Rarely have I gotten to visit sequencing labs, and the only time I've seen one with the cover open was when a Polonator was being shown off at a conference.  So it was a lot of fun to watch an almost hour-long YouTube video that Lex Nederbragt had pointed out in which someone who likes to disassemble complicated electronics performs his craft on a 454 FLX.

An interesting bit about this dissection is that the chap doing it has only a very limited prior understanding of what the machine does.  Far from being a problem, he provides a very entertaining commentary as he tries to figure out how the components worked together and why engineering choices were made.  Unfortunately, he has a slightly incomplete kit; the compute unit is missing which prevents him from trying to get some of the extracted subsystems from working.  So in many ways this is akin to an archaeology dig, trying to improve an incomplete understanding of a technology and the people who built it.  It really is fun to see this process of discovery.

I really had no proper perception of just how complicated an instrument these are.  Lots of custom electronics and custom machined elements.  His dryly humorous narrative points out a real zoo of different fastener types (though to his surprise, all metric, given this was designed in the United States).  The positioning hardware for the reagent sipper tubes is apparently something capable of very high precision and smoothness, which seems hardly necessary for the task at hand.  He points out the exterior finish is extremely good and speculates on how parts of the case were assembled.  

He also points out an odd order-of-assembly choice. The instrument is fastened to a piece of Corian sitting on a custom-built cart which holds the uninterruptible power supply.  Curiously, the placement of the mounting screws suggests the mating to the Corian occurred very early in the manufacturing process.

The dissector makes all sorts of interesting observations.  For example, a set of electromechanical interlocks make sure the PicoTiter plate door can't be lowered during a run, which would spoil the run.  But in case the power fails, there's yet another mechanical scheme to allow that to be overridden.  That's a lot of complexity to solve a single problem, but probably seen as necessary given the expense of accidentally terminating operations.

Seeing this video makes it much easier to understand why these instruments were so expensive.  A few parts, such as the high-end CCD camera in the 454, must have cost a pretty penny, but having so many components to assemble into subsystems and then assemble those into the final instrument would require a lot of labor.  It also brings home how Ion Torrent cut costs by eliminating the optical components.  Ditching the fluidics, as in the Oxford Nanopore and presumably whatever comes from Roche/Genia enables greatly simplifying the instrument.

There are some notes on cracking open Illumina sequencers out there as well.  A professor named Nick Kaplinsky at Swarthmore College, near where I grew up, has an extensive blog featuring articles looking at Illumina and SOLiD instruments.  Another fellow named Tito Jankowski has an extensive description of breaking down an Illumina GA.  I found those with a little looking; it wouldn't surprise me if there are more out there.

I really urge you to watch the video.  It is nearly an hour long, but well worth the time. It would really great if this fellow got his hands on the later 454 GS Junior instrument so we could compare the designs and see what was changed in that design.  Could also be fun to have him tear down one of each model of sequencer, comparing and contrasting design elements.  It would also be cool to have someone who really understood the instrument provide an overlaid commentary, to fill in a few mysteries, provide context and color, and make any needed corrections.


1 comment:

  1. 'reverse engineering', and internet postings thereof, usually breach license or T&C terms.

    ReplyDelete