[nickloman]

Yep, it does depend on companies not sharing serial numbers between unrelated product lines. Intuitively this feels unlikely to be the case but I guess it is possible.

We do have some data (e.g. from the map) which will help us figure out if the predictions are likely to be right, e.g. relative frequencies of 454s, Solexas and SOLiDs in the wild.

My boss always says "Why think, why not do the experiment?" - lets have a go and see if it works! Please send your numbers in!

[steven]

Have you heard of the 'Tank counting statistic'?

http://www.guardian.co.uk/world/2006/jul/20/secondworldwar.tvandradio

Assuming sequential serial numbers, you should be able to bound the number of machines 'in the wild'.

Right. We are assuming here that for instance, none of the companies uses the same series for several kinds of products.
Tank counting is ok as long as only tanks are numbered..

[nickloman]

Hi guys

As part of the high-throughput sequencing map site James Hadfield and I set up, I've added a little page to collect instrument serial numbers.

Once we've got enough numbers for any given platform I'll post the results on the page. Hopefully we can get enough serials for each platform to get definitive data on the numbers of shipped instruments.

This is entirely anonymous and I won't publish the serial numbers so you won't upset your instrument rep!

Go visit http://pathogenomics.bham.ac.uk/hts/serial to take part

Cheers!

Nick.

[dan]

Have you heard of the 'Tank counting statistic'?

http://www.guardian.co.uk/world/2006/jul/20/secondworldwar.tvandradio

Assuming sequential serial numbers, you should be able to bound the number of machines 'in the wild'.

[james hadfield]

Our GAs have serial numbers and apprximate installation dates of:
202 October 2007
350 May 2008
582 December 2008

[GW_OK]

Thank you all for your helpful feedback.

One more point: it looks as if 454 has been losing market share. Maybe that's why they just announced a "454 junior" with 1/5th the price and 1/10th the throughput of their current system?

I heard about the GS Jr. several years ago from a Roche rep (very hush-hush at the time) so I don't think rolling it out is in response to market share. Frankly I thought the project had been shelved until they announced it.

[cgb]

discounts

of course what we can never expect to see is how many machines are fully paid for, how many are discounted and how many are just placed for free.

[brjordan]

Thank you all for your helpful feedback.

So we are left with an estimate of 1,000 machines (total, ca. 50% Illumina, the rest shared between SOLiD & 454), which several people agreed with (this is also in line with a June '09 report by the Leerink Swann investment bank), and another much higher estimate according to which there would be more than 1,500 Illumina machines, and presumably around 1,000 others... The error bar is pretty large! My impression (for what it's worth) is that the CURRENT (dec. '09) figure is probably at or above 2,000 total since a) the Leerink report did not account for all genome centers (and much less all individual labs), b) there has apparently been a buying spree lately, and c) the evidence for 1,500 Illumina machines is anecdotal (serial numbers) but significant.

One more point: it looks as if 454 has been losing market share. Maybe that's why they just announced a "454 junior" with 1/5th the price and 1/10th the throughput of their current system?

All comments are of course welcome...

Bertrand

[dan]

Thanks Jonathan!

While Illumina is cheapest, you need a moderate to serious coverage for deNovo sequencing and so on...

Right, that is kind of what I meant by 'interpretation'. i.e. the difference between the cost of a 'finished base' vs. the cost of a 'raw base'.

I'll see how these numbers come out when I multiply through by my volume of data...

Dan.

[Jonathan]

Nah, maybe a little high!

I did some calculations for a small talk last year and based by numbers on the official pricing of the 'Institute for Genome Sciences & Policy' at Duke University, Durham, USA - last January or so:

Saenger: 150$ --> 96kbp in ~1000bp
Illumina: 1950$ --> 325Mbp in 2x36bp
454: 8000$ --> 30Mbp in ~260bp

Saenger: 0.0015 $ ~7bp/cent
454: 0.0004 $ ~ 25bp/cent
Illumina: 0.000006 $ ~ 1600bp/cent

unless I was having some serious miscalculations...or very off estimates...

While Illumina is cheapest, you need a moderate to serious coverage for deNovo sequencing and so on...

Best
-Jonathan

[dan]

On this subject... Anyone got good figures for 'cost per Mega base' for the Illumina, 454 and Sanger technologies?

I have the following numbers:

Sanger 5.00 pounds per megabase.
Roche 0.02 pounds per megabase.
Illumina 0.002 pounds per megabase.


However, they seem too low!

Can anyone interpret these values for me?

Can anyone offer other values with appropriate interpretation?

Cheers,

[GW_OK]

Our serial number is 1508. We got ours delivered in late September.

[brjordan]

Well, serial numbers are not always meaningful - remember the story of Hitler's party card number? (number 555 - but they had started the numbering at 500 so as not to give away their small numbers).
However if you saw #216 two years ago and now see #1599 - this looks pretty solid (not SOLiD!)...
Any other comments to confirm this? Recent serial numbers or other info?

Thanks to all,

Bertrand

[kmcarr]

If their serial numbers are meaningful then they have shipped over 1500. We just got an instrument a month ago and its serial number is 1599. Our original instrument, received about 2.5 years ago was serial number 216.

[brjordan]

This sounds very high (1,500 GA instruments shipped)... Are you sure they are referring exclusively to GA and not lumping together other instruments? Do you have any link to this statement?
Thanks,

Bertrand