$1000 genome has it been and gone?

[james hadfield]

Hi All,
At AGBT Elaine Mardis and Chad Nusbaum both presented on 500Gb chemistry on the Illumina platform. They were followed by Geoff Smith from Ilumina showcasing their internal 1Tb runs. Have we 'missed' the passing of the $1000 genome?

HiSeq at 1Tb is equivalent to 33 30x converage human genomes.
A run costs about £20,000 including a bit more than just consumables costs (this also makes my maths work out nicley).
This gives us £606 per genome.
A nice just under $1000.

What's the impact?

[nickloman]

What about 33 library preps? Probably only another £8k or so, but still needs to be factored in...

Still, I agree, we're basically there!

[wraithnot]

Hi All,
At AGBT Elaine Mardis and Chad Nusbaum both presented on 500Gb chemistry on the Illumina platform. They were followed by Geoff Smith from Ilumina showcasing their internal 1Tb runs. Have we 'missed' the passing of the $1000 genome?

HiSeq at 1Tb is equivalent to 33 30x converage human genomes.
A run costs about £20,000 including a bit more than just consumables costs (this also makes my maths work out nicley).
This gives us £606 per genome.
A nice just under $1000.

What's the impact?

I think there is a slight math error in your calculations. Unless I'm completely misunderstanding something, 1 Tb is about 333 1X 3-Gb human genomes or 11.1 30x human genomes.

The HiSeq 2000 press release quotes "...200 Gigabases (Gb) in a single run lasting eight days (paired 100 base reads) using dual flow cells." and that will "...drive the cost of a genome under $10,000". So I would say a 1 Tb run would equate to a $2,000 genome if nothing but the output changed. This is pretty close to what your calculations would yield if a full run generated 11 30x genomes.

But you also need to consider the cost of the machine in the calculations. If a HiSeq2000 costs $690,000, has an expected life of 3 years, and can perform 40 1 Tb runs per year, that works out to $5,750 per run. Or $522 per genome if I did my math correctly. So I would argue that a 1 Tb HiSeq 200 run would represent a $2,500 human genome if you assume that their internal run took that same amount of time and used the same reagents as a standard run. And this is ignoring things like the cost of labor, overhead, etc. Pretty amazing considering the cost of the first human genome. But not quite $1,000 yet.

[clivey]

Darn !

Wraithnot, you beat me to it - by minutes !!

Even if, in the unlikely event, it were true that it had gone and Illuminas marketing group had missed it -- I think that just shows how trite and irrelevant it has become.

[wraithnot]

Wraithnot, you beat me to it - by minutes !!

Even if, in the unlikely event, it were true that it had gone and Illuminas marketing group had missed it -- I think that just shows how trite and irrelevant it has become.

After a bit more digging, the original premise of this thread could be correct (at least if you limit yourselves to consumable costs). Here's the first paragraph of a press release from Complete Genomics:

MOUNTAIN VIEW, Calif. — Nov. 5, 2009 — Complete Genomics, a third-generation human genome sequencing company, today announced publication of a report in the journal Science describing its proprietary DNA sequencing platform, including analysis of sequence data from three complete human genomes. The consumables cost for these three genomes sequenced on the proof-of-principle genomic DNA nanoarrays ranged from $8,005 for 87x coverage to $1,726 for 45x coverage for the samples described in this report.

If your benchmark is 30x coverage and you extrapolate their $1,726 for 45x statement, you get $1,150 for a 30x genome. And that was 16 months ago. Of course there is a lot more to the real cost than just the consumables. And they need to make a profit on top of that to stay in business. So you won't be sending them a tube of DNA and a check for $1,000 and getting back a 30x genome any time soon.

[krobison]

For better or worse, when folks talk about "$1K" genome it tends to be just reagent costs for the sequence acquisition -- no labor, library prep, machine amortization, electricity, downstream computing, rent, postage, etc.

Reagent costs are nice and solid, whereas many of those others have some free parameters (what labor rate, rent, time to amortize, etc) which are up for argument, but of course the reagent-only scheme gives a number of very limited utility.

I think the X-Prize criteria do include all of the costs, with fixed values for the free parameters, but I've never stared at them.

[clivey]

No

There's cost of goods and price.

As a servce business, you can charge what you like when you have a capital cash pile in the 10s of millions. At least for a while. That cash pile is decreasing not increasing. This is not factored into the calculation. If it were priced at 1k, and the cash pile was increasing due to profits, that would be convincing.

[Richard Finney]

Collection, storage, real estate, labor, capital costs , etc. need to be factored in, not just the ingredients. And when it hits the clinic, there will be liability costs!!! ouch!