Estimating the number of sequencers in the wild using a WWII formula

i-f49c539b3e59fc2ab53de566a297a461-loman_seq_map_100319.jpgNick Loman (of the University of Birmingham, and the Pathogens: Genes and Genomes blog) has a post updating us on his survey of second-generation sequencing machines around the world. Loman's results are also available in the format of a handy Google map (see left).

The take-home messages based on 669 machines in the database: Illumina continues to utterly dominate the second-gen market, with competing short-read platform SOLiD squabbling for scraps with Roche's 454. That's a pretty poor outcome for SOLiD, which has failed to gain traction in the market despite having the full force of Life Technologies' marketing machinery behind it.
Meanwhile, Nick has also hatched a fiendish plan to refine the estimate the number of second-gen sequencing machines in the wild, using an approach pioneered by British statisticians seeking to estimate German tank production during World War II. 
The strategy is simple: by recording the serial numbers of sequencing machines and the date on which they were received, and making some assumptions (e.g. the serial numbers are in fact continuous), one can generate a probabilistic estimate of the rate at which machines are being distributed.
There are certainly a number of ways in which such estimates can prove inaccurate - but as Loman says, it's still worth doing the experiment. So if you're in charge of second-gen sequencers, add your serial numbers now.

More like this

Thanks for the links Daniel! It is interesting that you pick up on the poor performance of Life Technologies. I tend to think they've done well to sell as many machines as they have.

Certainly they seem to have sold nearly as many machines as 454 which I find surprising. You can make a strong case for getting a 454 as a complementary technology because it does certain applications very well - i.e. the ones that really benefit from the longer reads (de novo assembly, 16S metagenomics, cDNA sequencing etc.).

But I fail to see why someone would have bought a SOLiD over an Illumina, at least on pure technical grounds. Probably now the machines are similarly easy to use and produce comparable throughputs, but this wasn't always the case.

I guess you are right that they've put in some serious marketing effort.

Keep up the great work on the blog


Hey Nick,

I commented on SOLiD in that way simply because they tend to frame themselves as direct competitors to Illumina (i.e. short-read, high-yield platforms) and because (as you've commented yourself) they seem implausibly perplexed by their second-place status behind Illumina.

But I completely agree with you that there's no sound technical reason why someone would have purchased SOLiD over Illumina, and of course it's no coincidence that virtually all the major genome facilities have ended up with wall-to-wall Illumina machines. Perhaps you're right: I should be commending SOLiD on the remarkable marketing success of selling any machines at all!

I think a key part of the MI5 plan was not telling the Jerrys what they were doing.

By dontworry (not verified) on 19 Mar 2010 #permalink

OK Daniel, this time your british humor came through well. Per dontworry, marketing groups are now going to go the production groups and make them put new serial number stickers on the machines to wreck Nick's plan. It's great to see this data.

Thanks for the post, pretty interesting stuff. It will be really interesting to track the deployment of the next round of machines, including the smaller machines (like GA IIe, etc.) single molecule sequencing, etc. As others have commented ( the sequencing instrumentation industry will likely go through fragmentation moving forward. It'll be really cool to watch this in real-time.

Do you know what the delay would be in obtaining such data?