is sequencing cost still dropping?

[xyzzy123]

Looking at the graphs at

https://www.genome.gov/sequencingcosts/

the cost for 2012 has actually gone UP a bit near the end of the year (very little, to be sure, but it's effectively flat for the entire year, after plummeting in the previous 4 years). I'd expected that when the cost got low enough, more emphasis would be placed on accuracy, slowing the drop. I suppose it's also possible that when the cost gets low enough that insurance pays for it, there would be no further incentive to cut price. Can someone more knowledgeable explain what's likely going on here?

[JPC]

With their market dominance my initial thought is to blame Illumina for the increase in cost shown in this graph (I think their reagents costs went up early 2012 ?) but other factors would certainly be significant. Possibly a growth in bench top sequencers has had an affect, these are not optimal cost wise but they sure are popular.

I would have add the disclaimer that my groups seq costs have not increased over 2012 so my point of view is not from a particularly informed position.

[xyzzy123]

Another data point was just posted for early 2013 and it appears the cost is finally starting to drop again.

[xyzzy123]

Another data point was just posted and the cost over the last 1 1/2 years appears almost flat, not even dropping at Moore's Law rate any more.

[scbaker]

Another data point was just posted and the cost over the last 1 1/2 years appears almost flat, not even dropping at Moore's Law rate any more.

The lack of dropping costs is purely due to market forces. I explain it in more detail at my blog: http://biotechmarketer.blogspot.com/...ising-ngs.html

tl;dr version = Illumina simply isn't seeing the competition that would necessitate them to drive down the costs. They (and the buyers of NGS technology) are focusing on other factors (e.g., speed).

[xyzzy123]

Thanks for that (and also the link to Mick Watson's blog post at the end). I'd like to see non-medical applications take off (for example ancestry) but that won't happen unless the cost drops by at least another one or two orders of magnitude.

Edit: I read your link to Neil Hall's "After the Gold Rush" and thought it was ironic that he wrote "even if Oxford Nanopore (or someone) were to deliver a revolutionary technology this year, the crunch will happen soon." in a medium that only exists because Moore's Law has continued for over 50 years (longer if you count the era before ICs). Sure, all exponential growth stops eventually, but "eventually" and "soon" are not the same. I also think his belief that the end of the cost drop would be a good thing is pretty naive considering that without the same type of extended drop in computing costs, sequencing something as large as a human genome wouldn't be possible at all, much less cheaply. Pencil and paper won't cut it.

[scbaker]

Sure, all exponential growth stops eventually, but "eventually" and "soon" are not the same.

Agreed. The price drop WILL stop, but I think it will make a huge difference whether it stops at $1000, $100 or $10 for a whole human genome. Same goes for the ag market you mentioned - lots of samples waiting to be sequenced for the right price.

Edit: for some reason I read 'ancestry' as 'agriculture' in your post - not sure what I was thinking. But I still stick with ag as having a bigger potential than ancestry. And I also think other 'omes could be bigger than the genome - transcriptomes and epigenomes will need to be measured multiple times.

[scbaker]

I also think his belief that the end of the cost drop would be a good thing is pretty naive...

While I'm expecting and eagerly awaiting continued price drops, Neil has a point about the 'laziness' of some (most?) NGS experiments. There have been some pretty weak experimental designs where the researcher was clearly hoping the "magic" of NGS would save the day. I think this will change for the better as the technology matures.

[xyzzy123]

The next data point was finally posted, and it looks like it might be starting to drop at roughly Moore's Law rate again. Even if so, at that rate it would take roughly 15 years to get down to $10. I hope it's not that slow.

[ctseto]

There's also a competing market force push to increase read-length, though the increase of read density on the v3 flowcell might also indirectly contribute to reduced costs.

Of course, there's associated informatic costs that don't come into play when you look at the sequencing costs...

[xyzzy123]

The data point that was added 3 months ago was removed a few days afterwards. The data point for January hasn't been added either, so I don't know if the page is being updated anymore. Illumina claims to have reached the $1,000 genome recently with the HiSeq X Ten, though it's disputed whether that will be the actual cost. It would be the first significant drop in cost in 2 years.

Edit: The data points for July and October 2013 have been added, showing a very slow continued drop, less than Moore's Law (I confirmed the dates by looking at sequencing_costs.xlsx). Presumably Illumina's announcement will be reflected in the next point.

[xyzzy123]

Informative blog post regarding Illumina's announcement:

http://biomickwatson.wordpress.com/2014/01/15/illumina-destroy-the-opposition-again-almost/

[AllSeq]

Informative blog post regarding Illumina's announcement:

http://biomickwatson.wordpress.com/2...-again-almost/

There have been a number of interesting posts about Illumina's recent announcements - Illumina really grabbed people's attention!

We've written up a couple of posts on our blog, one about what it really takes to achieve the '$1000 genome' and another on how Illumina's new systems fit into the broader market (and what systems have been quietly discontinued!)

[xyzzy123]

Two more data points have been posted at https://www.genome.gov/sequencingcosts/ (January and April 2014) and though there is a significant drop in January, April reverses that. The trend of a much slower than Moore's Law drop continues.

[AllSeq]

Two more data points have been posted at https://www.genome.gov/sequencingcosts/ (January and April 2014) and though there is a significant drop in January, April reverses that. The trend of a much slower than Moore's Law drop continues.

AT $4k per genome, these calculations appear to NOT be taking Illumina's new HiSeq X Ten pricing into account. 30X Human WGS is available through AllSeq's marketplace at high volumes for as little as $1450 (for just sequencing - $100 more each for library prep and sequencing alignment). Even at lower volumes the price generally isn't more than $2000. But as these prices are only for human WGS, maybe their exclusion from the cost chart is justifiable.