Is there any empirical estimate of idle sequencer capacity available?

[DeeDeeBSE]

At my local university I have heard rumors that there "is a high percentage of idle sequencer capacity worldwide." Unfortunately I was not able to directly ask the person making the claim where he got his information, so I'm trying to track down more information about that statement. Some basic Google searches haven't turned up anything interesting, so I'm hoping someone here can provide more perspective.

[kopi-o]

BlueSeq (http://www.blueseq.com/) should be in a position to know. Perhaps you can ask them.

[scbaker]

Thanks for the mention, kopi-o!

This is a really interesting question. We’re working on tools that will allow us to estimate things like this in the future, but for now we can make a ‘back of the envelope’ calculation.

Let’s just look at HiSeqs (which would account for the majority of the world’s sequencing capacity). Illumina has stated that each HiSeq generates ~$300k in consumable revenue per year. If we assume each full HiSeq run costs $20k, then this translates to 15 runs per year. I would say the theoretical maximum runs would be 25/year (in the ‘high throughput’ mode). Therefore, HiSeqs are being run at ~60% of the maximum (15/25). 10 runs times 600Gb per run = 6Tb excess capacity per machine. Multiply that by the number of machines (let’s say 1000 to make the math easy) and you have 6 petabases of excess capacity per year.

If we do the same thing for MiSeqs…
$50k/year in consumables = 50 runs = ~33% of maximum capacity (3 per week).
1000 machines X 100 runs X 10Gb/run = 1Pb excess capacity per year.

I can’t find the stated revenue/year number for the PGM and Proton, but if you have that you could make similar calculations. Let’s just say they contribute another 1Pb of unused capacity/year, giving us a (very rough) total of 8Pb. Lots of factors could bump this number up (more instruments/higher throughputs) or down (theoretical max estimate not conservative enough), but it’s probably in the ballpark.

Cheers,

Shawn

[GenoMax]

There is no doubt that unused sequencing capacity exists but there are two facets to that question. For consumers: how to tap into it productively and simply? For providers: How to accurately estimate the spare capacity and make it available "just in time"?

At many universities items like sequencers are purchased from capital expenditure funds meant to boost infrastructure capacity for internal needs. As a result internal samples always have priority. We run a mix of several HiSeq/MiSeq and there are (rare) times when a machine or two idles for a length of time. We had thought about using that time but since we can't accurately predict the availability it was not very practical.

Most commercial customers are looking for sustained availability/capacity over a length of defined time, on the other hand dealing with small infrequent samples is not worth the logistics involved in handling/billing/delivery of data.

[mbblack]

To add to what GenoMax mentioned, a lot of Universities bought and buy large capital equipment like sequencers without any strategic plan in place to actually make use of it. That's because traditionally, it has been simple for an individual PI to get a strategic funds grant to buy a piece of equipment without running it through anyone else at the institution to plan out operational logistics or costs.

So, I would hazard a guess (one I'm pretty confident on) that the vast majority of academic sequencers are not used to anywhere near their actual capacity, many likely never even come close to enough use to really justify their capital cost in the first place. In other words, funds would have been better spent assisting investigators to get their sequencing done elsewhere, rather then the expenditure of funds on an under-used local facility.

Obviously, I'm painting with a pretty broad brush, but I worked in academia for a long time, and I don't think I'm far off for many institutions (although there are definitely some exceptions out there). And the same thing can be said for a lot of other high cost capital equipment at Universities for a number of reasons (but we don't need to go there).

[DeeDeeBSE]

Thanks for the mention, kopi-o!

This is a really interesting question. We’re working on tools that will allow us to estimate things like this in the future, but for now we can make a ‘back of the envelope’ calculation.

Let’s just look at HiSeqs (which would account for the majority of the world’s sequencing capacity). Illumina has stated that each HiSeq generates ~$300k in consumable revenue per year. If we assume each full HiSeq run costs $20k, then this translates to 15 runs per year. <snip>

I like this approach! Thanks for your input!

[scbaker]

...So, I would hazard a guess (one I'm pretty confident on) that the vast majority of academic sequencers are not used to anywhere near their actual capacity, many likely never even come close to enough use to really justify their capital cost in the first place....

Yeah, that definitely meshes with what I've seen. But given that, it's pretty remarkable that HiSeq's are running ~60% of full capacity (where 100% probably isn't even achievable as all machines have some downtime). MiSeqs and Protons/PGMs are being used for smaller/quicker experiments (with more relative hands on time/overhead per unit of output), so it's not too surprising to see them being used more like 30% of capacity. My guess is that SOLiD machines are running at very low utilization rates (with many just gathering dust). 454s may be somewhere in between.

[rskr]

To add to what GenoMax mentioned, a lot of Universities bought and buy large capital equipment like sequencers without any strategic plan in place to actually make use of it. That's because traditionally, it has been simple for an individual PI to get a strategic funds grant to buy a piece of equipment without running it through anyone else at the institution to plan out operational logistics or costs.

So, I would hazard a guess (one I'm pretty confident on) that the vast majority of academic sequencers are not used to anywhere near their actual capacity, many likely never even come close to enough use to really justify their capital cost in the first place. In other words, funds would have been better spent assisting investigators to get their sequencing done elsewhere, rather then the expenditure of funds on an under-used local facility.

Obviously, I'm painting with a pretty broad brush, but I worked in academia for a long time, and I don't think I'm far off for many institutions (although there are definitely some exceptions out there). And the same thing can be said for a lot of other high cost capital equipment at Universities for a number of reasons (but we don't need to go there).

It is kind of like the high throughput computation when they need it they need it done now and don't want to have a whole team of scientists waiting idly for the computer/sequencer to finish, so they buy excess capacity expecting to use it for certain key experiments, nor do they want to spend the time thinking about developing protocols to do it with more modest resources. Also they get political points for having a sequencer/high throughput computation facility and prevent internal institution competition from making progress, since an institution that barely uses one can't very well use two.