DNA Technology Giant Illumina Set to Move Whole Genome Sequencing Down to $200 Price Point in 2023

Question

I've included mention of this development in other G2G posts the past couple of days (read: I've buried it amidst typically overdone word-count), but I think it might deserve its own thread.

I wrote about the new Illumina NovaSeq X series of DNA sequencers on September 29, the day the press releases made the rounds: https://countingchromosomes.com/blog/94-illumina-unveils-the-new-novaseq-x-series-of-dna-sequencers.

While we aren't likely to see a significant near-term impact to our use of DNA for genealogy, that the sequencing technologies continue to make rapid leaps forward will eventually begin to alter the playing field for genetic genealogy. Last year saw the first ever full sequencing of a human genome by the Telomere to Telomere Consortium using hybrid long-read methodologies from Oxford Nanopore and Pacific Biosciences. These techniques allowed us to see approximately 7% of the human genome never before sequenced.

Illumina's new announcement--while using more traditional short-read sequencing--promises a perhaps broader effect by making testing throughput up to 2.6 times faster and lowering the anticipated cost, per sequencing, to around $200. Illumina possesses approximately 80% of the global market share for DNA sequencing, and the new systems will begin shipping after the first of next year.

Putting the potential price reduction in perspective, in 2011 it cost approximately US$10,000 to sequence a human genome. By 2018 the estimated baseline was about $1,000, and today the number hovers around $600.

By contrast, our common autosomal DNA tests for genealogy are still using both techniques and genome reference maps from 2012. Improvements have been made since then, of course, but nothing more significant than shifting around which SNPs, or markers, we test, and the models (e.g., genotype imputation, computational phasing) we use to interpret the data.

Genealogy, as much as we might like it to be otherwise, represents only a small fraction of the economics of DNA. Medical and pharmaceutical uses dominate the flow of money, and that's where the new sequencing systems will be felt first. It will involve significant capital outlay and investment in new processes, workflows, training, and computer infrastructure and algorithms for genealogy testing companies to finally make the leap into the 2020s.

Too, we've somewhat backed ourselves into a corner regarding tests taken previously and compatibility/matching with whole genome sequencing results moving forward.

However, progress waits for no one. Either genealogy testing and reporting services will begin to adapt to a technological environment where we can sequence a whole genome about as quickly as we can do the lab processing for our low-density microarrays today--and perform that sequencing for only a little more than twice the current $79 average for a microarray test--or DNA testing as we currently know it will go the way of the fax machine in a matter of a few years.

Edited: Corrected a copy-and-pasted duplicate word. Darned tyops...

Comments

I think the major problem with microarrays is that we must deal with unphased data. WGS won't necessarily help with that. I also think the added information content of a WGS is somewhat marginal, since most people will match most people on the bases between the SNPs. What I do think could prove interesting is identifying novel, familial mutations, like the one that accounts for the hearing loss that runs in my family. I've traced it back to 1813, yet so far no one who matches us on the "segment" has that mutation. It's a bit like mtDNA, where an intermediate level of testing at 23andMe can give you a haplogroup but not a haplotype.

The actual accuracy of microarrays is excellent for base calls, and I think perhaps even better than WGS. I have done microarray and WGS for my sister and myself, and the segments for the chromosome browser on GEDmatch look much cleaner for the microarrays than the WGS.

Answer 1

Great news, and thanks for sharing!

The day that a hobbyist can get a DNA testing chip and hook it up to an Arduino to make a home grown testing lab, then I'll be blown away. ;-)

Comments

 We're still a while away from do-it-yourself PC microcontroller sequencers. The new NovaSeq X Plus, the first of the two models that will ship, has an estimated price tag of $1.3 million. That may come down some once they actually begin delivery, so you might be able to pick one up for a mere $1.1 or $1.2 million.

Joking aside, if you aren't aiming for that level of functionality there are, well, "microlab" options to play with in a home basement. They won't be able to yield the kind of results you can use for genealogy--no sequencing or microarray hybridization; more of a PCR plus gel electrophoresis approach--but they do exist. With some of them you can do minimalist Sanger sequencing, too.

One option is the "Bento Lab" from Bento Bioworks, https://bento.bio. Their entry-level package is about US$1,600; the "Pro" version is about $2,000. They also sell supplies like reagents, buffers, pipettes, and PCR and centrifuge tubes in small quantities for the do-it-yourself geneticist.

I used one of these in 2020. Pretty neat, but it didn't motivate me to run out and buy my own. It can't deliver the kind of data I'm interested in. Would make an interesting conversation piece when the family gathers for the holidays, though. 

---

Let's give it another 10-15 years and see where the tech is at then. ;-)

---

A la Moore's Law. Yep; 10 to 15 years is like 60 to 90 years in dog-years and DNA-years. 

---

A computer generation is every Pi years (or half Tau, if you want to be pedantic). Give it 4 generations then. ;-)

---

(That sound is the gears turning) Ever wonder if you're related to King Tut? Well now you can know with the new DNAMaster 5000! Want to know if your dog is related to the one you had as a kid? Or is your dog related to you? Or why you like coffee so much?  The DNAMaster 5000 has you covered! It's as simple as spitting into one of the DNAMaster's receptacles and in just five minutes, you will know if Genghis Khan was one of your ancestors. Want to know how you relate to someone else? Both of you can now spit into the DNAMaster's two receptacles to find that your common ancestor was Atilla the Hun or Boudica!

You too can own the DNAMaster 5000 for only a first payment of $19.95 (followed by 60,150 monthly payments)! Order yours today!!! 

Answer 2

While all news is good news (especially if it gets discussion going), I think a more balanced coverage would be helpful here.  A lot already given is from statements by Illumina in their recent press releases. Which are very slanted; just like information and releases from a well known genetic genealogy company who shall remain nameless here ;)

Let me address two aspects.  Use of whole genome sequencing in Genetic Genealogy.  And the price point of whole genome sequencing.

(1) A significant and growing number have been using whole genome sequencing in genetic genealogy.  I personally have only done such testing since 2019.  With over 30 under my belt directly. WGS is a reality although not a direct one in with many sites today. There is no question of the advancement it gives in true whole sequencing of the Y and mitochondria; in doing phylogenetic tree of haplogroup work.  WGS regularly delivers far greater results than the latest and best BigY-700. Hands down. See https://h600.org/wiki/Y+DNA+Sequencing+Comparisons for one example. The Y-DNA Warehouse has more.  To date, there is not a great advantage in the increased accuracy and coverage of the autosomes for segment matching sites.  The biggest benefit you get is in being able to generate microarray results for all available test kits on the market.  When submitting such to GEDMatch and Geneanet, for example, you have a much greater use of real SNPs and not any assumed or imputed ones.  At minimum, you can generally upload to any site a 23andMe v3 and v5 merged result test kit which includes around 1.2 million SNPs.

The only other testing I consider is Ancestry and 23andMe to get to their closed match databases. They have, by orders of magnitude, the largest match databases. And that is what is important in genetic genealogy after all.  (I might add they are about the only vendors that deliver the full microarray results which, by design and default, test all the DNA including the Y chromosome, mitochondria and X.  Other companies strip those results out before delivering.)

(2) Since 2019, the WGS has been available for $200-300 on a regular basis.  The vast majority of testing in the genetic genealogy community has been with Dante Labs.  Who are using Illumina Novaseq 6000 sequencers.  I will not get into the arguments of whether one can sequence for that cost with Illumina here (you can, they do), suffice it to say the service is available to the consumer and has been for several years.  Nebula Genomics is the next big tester and they use BGI's lab that uses their subsidiary MGI for equipment.  MGI actually delivers comparable and often superiour results (especially at the cost per gigabyte) than Illumina. Their main technology is from a startup in San Jose, CA they purchased a number of years ago.  

MGI announced in late 2019 that they were close to achieving the $100 human WGS cost. Just last year at this time, Illumina announced plans to achieve a $600 human WGS cost. No one will argue that Illumina is the sequencing industry behemoth. But they have consistently been falling behind in new R&D.  Although they spend large amounts on R&D, a majority of their advancements appear to come from acquisitions.  IMHO, there are two more recent items that is causing this pre-announcement of capability by Illumina. And less than a year after announcing their plan to achieve a $600 human WGS test cost (that has not been delivered to all customers yet).

First, BGI has achieved and surpassed the accuracy and low cost.  Their DNBSeq T10 is a massive machine with automation that is used in the large BGI lab (and regularly by Nebula). So they are achieving the $100 WGS through large volumes and scale.  Ultima Genomics says they will do it with desktop systems. Illumina's announcement claims they hope to get to $200 sometime in the future with a similar desktop system and new, advanced library prep chemicals. Illumina successfully stopped MGI sales in the USA and a few other locations through patent infringement lawsuits. All those supposed violated patents have now expired and MGI is now availabe in the USA market as of last month.

Secondly, the addition of credible, heavily-funded, new alternatives that are going after a lower cost per sequence. That is, Element Biosciences and Ultima Genomics being the two largest I am thinking of. They have been getting a lot of funding, favorable reviews and begun limited testing with key customers and collaborators.

Now Microarray testing has been around awhile for both medical and genealogy uses.  And thus has a great economy of scale already achieved that can still be utilized at a low cost for continued sales.  But the advantages of the current NGS (short read, shotgun sequencing) is clear and costs are way down to favorbly compete for use.  And likely to replace microarray within the next five years.

Personally, I am there already and looking forward to the next available technology.  Specifically, the PacBio HiFi CCS sequencer used by the T2T and Human PanGenome Project to get that complete human genome sequence.  Its key advantage being a highly reliable, long-read sequencing technology. As reliable in reads as current NGS short read sequencers (99.9%).  But with read lengths of 25,000 base pairs (versus 300-400 from short-read sequencers).  Nanopore, which does long read also, is still several orders of magnitude awayin realiability of Illumina, MGI and PacBio.  Long read really helps to better recreate what you actually see in an individuals genome.  So more complex variations than just single-base-pair and short repeats can be determined.  There is a hope that this test technology will be available at a cost of $1000 per human WGS in the coming year.

EDITED: I had missed the blog post reference at the top.  That is much more balanced and revealing.  But still not really covering the proposed product versus actual from other vendors. I know Illumina tries to take a page from the Apple playbook. But like they used to say in the business world for 30+ years -- you never got fired if you bought IBM ...

Comments

And for some additional information.  I believe there are well over 50,000 Consumer (under $500) WGS tests so far from the 4 main suppliers: Dante Labs, Nebula Genomics, Sequencing.com and ySeq.net.  Dante leads with 35,000 to date.  Less than half appear to be genealogy-solely focused (or maybe better said "looking solely for health analysis" from the result). I have not done a literature search, but some in the Ancient DNA community, like David Reich's lab at Harvard, estimate over 25,000 WGS runs of remains. Many such results have been posted to yFull's Y and Mito trees.

For about 8 weeks in August / September, Dante Labs had their annual late summer sale for $200.  All orders in the US were fulfilled by GenebyGene. Kits sent out by and returned to their Houston lab and even sequenced by them.  Bioinformatics appears done by Dante Labs AWS servers using their standard Illumina DRAGEN pipeline.  Results were almost always returned within 2 weeks after receipt of the kits in Houston.  The sequencer serial number shows it is the same Illumina Novaseq 6000 being used for BigY-700 tests. Dante has 3 in their Italy lab and we believe more in their new labs at other locations such as the Middle East. We estimate between 1-3 thousand kits were sold during that period; both in the USA and the rest of world handled by Dante Labs in Italy.

1 August was also the start of a switch over by Dante from spit tubes to blood sampling kits. Non-USA customers get the Tasso M20 kit which fills a few, very small capillary tubes which are then "dried" out for shipment.  So not as time or temperature sensitive.  Kits from GenebyGene used the Tasso Plus (which has just gone off the market already; presumably replaced with the SST). The Plus / SST retrieves a vial of blood that must be capped and sent back.  The blood sampling appears to have much lower contamination rates in returned results than most Illumina WGS processing runs we have seen to date.  Spit tubes and swabs are still available upon request if made after ordering and before shipping.

I have to admit, it is a little harder to get a relative to use the blood sampling method than the spit tubes or cheek swabs.  Especially if you are not there to do it and assure them.  But if it increases the reliability, especially with older testers. So something we may have to learn to live with.

---

Excellent answer, Randy.

How does the new blood based test work vs the spit-in-the-tube / scratch-the-cheek?

Is it similar to a diabetes test or what amount of blood is taken for Dante’s WGS?

It was the only thing that stopped me from ordering one for my mum with her being 12h by plane away.

---

Andreas, the blood sample appears to be more reliable (for Dante at least).  We are regularly seeing map rates of 98+ % from both Italy and Houston on the Illumina sequencers. So I suspect we will get less lab contamination issues than in the past with the saliva / cheek swabs at Dante.  But Thomas at ySeq would be a good one to answer.  He uniquely uses dry cheek swabs reliably. Which are very easy to "ship" (actually, can be dropped in a regular envelope and sent regular mail across borders).

Note that the Tasso M20 is used by Dante everywhere outside the USA.  It is a dried blood sample so easily ships across borders and without special classification.  The USA Dante orders (from GenebyGene) are using the Tasso Plus (maybe SST now) that is a vial of liquid blood.  So more shipping restrictions and a need to get it to the lab without delay.  But we have heard of no failures as of yet.  GenebyGene is using the same USPS labeling for return shipment as they do for FTDNA sample kits.  But getting the delivery direct to the lab (as opposed to a clearing warehouse like for FTDNA samples that takes another week or so to clear).

98%+ mapping of WGS is pretty regular with the BGI lab used at Nebula / Sequencing.com even though they use wet swabs like FTDNA and MyHeritage.  A lot has to do with the library prep process to try and remove bacteria from the sample. Either MGIs is superior over Illumina or the package used and handling is poorer at Dante.  Maybe a combination of both. ySeq has an MGI machine but a desktop version instead of the large, multi box T7/T10 that BGI uses. Novaseq 6000 are desktop as well.

Whether dried or liquid blood sample though, the Tasso device is a little tougher to use and does cause some minor pain to the arm (like a needle for a regular blood draw might).  So am concerned for its use with elderly relatives; especially ones on their own. Which seems to be common.

The Tasso M20 takes about 75uL of blood while the Plus takes 200-300 uL.  I think the SST is half that at 150 uL max. The M20 draws it into capillary tubes and chambers. Then has you remove plastic covering which allows it to air dry.  The Plus / SST requires you remove the vial from the device and put on a cap. Which is possibly more prone to spillage and maybe even contamination. All Tasso solutions use the same device that attaches to the arm (via sticky tape) and then slightly punctures the arm and uses a slight vacuum action to then draw blood out. Some feel the prick. Some have difficulty pressing the button to start the action -- especially if trying to do it on their own as you cannot see your upper arm where you are trying to attach and press.  Not sure how much a diabetes test strip draws but suspect below that 75 uL range.

The Tasso devices come with short instruction cards as well as QR code links to online videos showing its use. Unfortunately, this material has been pulled for the Plus devices.  See https://www.tassoinc.com/ and select the device of interest. The Plus video is still available with others at https://www.tassoinc.com/product-use-videos . Not sure if the plus change in availability is due to updates because of recent FDA full clearance or they are now stopped from providing that product.  Technically, the Plus is only supposed to be available in the USA with a prescription (Rx) and is labeled as such. An interaction with Tasso customer support indicated you have to talk to the lab who bought the device about instructions.  Got the feeling they are simply restricting info to licensed labs now.  The Plus does use the same tube and color-coded top that labs use when doing blood samples.  Not sure about any restrictions with the M20 or SST.

UPDATE: Found the Tasso Plus kit page on archive.org: https://web.archive.org/web/20210829124926/https://www.tassoinc.com/tasso-plus  They are shipped with the lavender cap which is a 300-500 uL sample size. IThe tabs are shown linearly down and you can click on the English instructions PNG and it will come back with a PDF after some redirection.