I took a look inside Complete Genomics' lab. What I saw turned me into a believer.

The future of human genetics will arrive when we understand how each gene in your DNA interacts with every other gene to form who you are. In order to get that understanding we’re going to need to sequence a huge number of genomes. And in order to sequence genomes we’re going to need Complete Genomics. The Mountain View based startup recently went public (NASDAQ: GNOM); with their requisite silent period ended we finally got a chance to tour their facilities and talk again with their CEO Cliff Reid. We were impressed. Complete Genomics isn’t going to conquer the entire world of DNA testing, but the kingdom of human genome sequencing they’ve carved out for themselves looks amazingly well fortified. As far as I can tell, if CG really takes off no one else will be able to challenge them in whole human genome sequencing for years to come. We asked Cliff Reid about the IPO, the genomics market, and some recent advancements in Complete Genomics’ analytical software. Check out his answers in the videos below.

Complete Genomics doesn’t sell a device, they sell a service. Companies like Illumina make DNA sequencers they sell to other companies. Reid and his gang take DNA samples and return with huge amounts of DNA data. Customers never need to deal with the biochemistry. That full service approach helps Complete Genomics focus on increasing the number of human genomes they can sequence for customers, and lowering the price for each. In the following clip, Reid gives us an update on the company, including the success of the IPO, and a look at its production capability and the current price of sequencing human genomes.

Getting to explore Complete Genomics’ Mountain View facilities was pretty amazing. I was struck by the relatively small size of the sequencing lab (though there’s plenty of room for expansion) and the degree to which automation is used. The company can produce 400 whole genome sequences each month, but there’s only ever two technicians working at any given time in the sequencing room. Robotic instrumentation and equipment makes the lab almost seem like a ghost town, but there’s a lot of action going on. I wish we could have turned our cameras on while touring that place, I bet many of you (and all of CG’s competitors) would have loved to see what they had cooking.

While we didn’t get any trade secrets out of Reid, the recipe for Complete Genomics’ secret sauce is pretty clear. This company is about doing one thing and doing it really well. Other big names in sequencing (Illumina, BGI, etc) are looking to handle many different kinds of genetic reading such as examining RNA, sequencing microbes, and sequencing animals. Complete Genomics is doing just one thing – whole human genomes. With that laser-like focus comes the ability to scale their entire company around a central mission. As we’ve said in previous coverage of this company, economies of scale, and single-mindedness in action will help Complete Genomics get cheaper and cheaper, faster and faster.

Eventually that means you’ll have access to inexpensive sequencing for your own genome.

Yet it will take a while to get there. Most of the whole genome sequencing done today is purchased by research labs, not by individuals looking to understand their own health. That’s how it should be. It’s going to take years for researchers to fully understand our DNA and what the variations in our genes mean. Selling genome information to individual customers like you and me would be premature. For now, Complete Genomics is focusing on the research market and helping them make the discoveries that will revolutionize genetics. Once the price per genome comes down, and the science behind genetics is better understood, Complete Genomics will transition into the consumer market. Reid says as much in the following video:

Complete Genomics isn’t 23andMe, Pathway Genomics, or any of the other companies selling genetic tests on the market today. First off, those groups are only testing a small subset of your entire genome for simple variations in individual letters of DNA (single nucleotide polymorphisms or SNPs). Complete Genomics is sequencing whole genomes – all 3 billion letters of DNA. Just as importantly Complete Genomics is waiting until researchers know more about the genome before they try to offer genome sequencing to consumers. That’s different from the current genetic testing companies that the FDA and congress are worried about. Those businesses will sell you genetic tests today, while much of the science is still not completely understood. Don’t get me wrong, I think there’s a place for that. Complete Genomics, however, is waiting until there’s an actual use for you to know your whole genome sequence before they try to sell it to you.

In the years ahead someone is going to have to process thousands (millions?) of genomes before we understand our DNA well enough to make the sort of pinpoint health recommendations that the public expects from this promising technology. Who’s going to sequence all that DNA? My guess is Complete Genomics. Again, getting back to that focus on whole genome sequencing, the company has built it’s entire technology around reading human DNA faster, better, and cheaper than anyone else.

The center of their business is their fluorescent read technology, which analyzes nanoballs of DNA that correspond to different A, C, T, G ‘letters’ of genetic code (learn more about the nanoballs on the Complete Genomics website). In Reid’s words they, “use silicon to read silicon”. A silicon based semi-conductor camera is precisely aligned with a different silicon wafer full of tiny pockets where the nanoballs sit. When I say tiny, I mean tiny – those pockets are just 300 nanometers across. The camera only has to use a few pixels (soon even fewer will be needed) to optically read each nanoball and determine if it contains an A, C, T or G.

According to Reid, Complete Genomics had reached the limits of optical reading technology. They are (or soon will be) using as little digital information to read each letter of DNA as you possibly could. The best other companies could probably do is match them. Add in the fact that their automated reading technologies are getting faster, and their analytical software is getting smarter, and you can see that Complete Genomics is quickly making itself the foremost leader in whole human genome sequencing.

Even if another company finds a way to sequence whole genomes with the same speed as Complete Genomics, I still think Reid and his cohort will come out on top. By focusing on just one product (whole human genomes) Complete Genomics will develop an unprecedented amount of experience in their field. They’ll discover all the little undiscovered idiosyncrasies that make human genomes unique compared to other sequencing projects and they’ll understand those twists and turns better because human genomes will be the only thing they work with. Single minded focus will help Complete Genomics learn all the tricks, and this will help make them more streamlined, faster, cheaper, and better at analyzing human genomes.

In the following video, Reid discusses Complete Genomics’ technology. Recently, the company announced a particularly exciting advancement in their software for digital analysis of DNA. Now when a researcher sends a genome to be sequenced CG will include information about copy number variation and structural variation – data that can illuminate where in a cancer cell genome that DNA has gone bad. Basically Complete Genomics is making it much easier for scientists to translate genomic data into insights about the nature of cancer. Coupled with a growing assortment of open source analytical code (CGA tools), Complete Genomics’ software is accelerating research into genetics. Pretty cool.

Other companies, like Ion Torrent, have developed DNA reading technologies that use semiconductors rather than optics to sequence genes. These CMOS based sequencing technologies have been hailed as the next step forward in genetics. As you saw in the video, Reid explained the reason why they aren’t afraid of semi-conductor based DNA readers. Can we all say that reason together? Complete Genomics only has to worry about whole human genome sequencing. While we toured the Complete Genomics facility, Reid pointed out that if you calculate how much current CMOS tech will cost to sequence a full human genome, the price would be many hundreds of thousands of dollars (or more). The big advantage semiconductor sequencing has is that it can fit on a desktop. Complete Genomics doesn’t want to fit on a desktop and they don’t want to be portable. They just want to sequence whole human genomes.

The same logic applies to those companies looking to push the boundaries of how much DNA they can read at a time. In the sequencing field, there are both ‘long reads’ and ‘short reads’. The human genome has more than 3 billion base pairs (those A,C,T, G letters). In order to sequence a genome you end up first sequencing smaller chunks of those letters, and then pasting those chunks together to get the full 3 billion letter code. The shorter the chunks you use, the more you need to double check your work and the more complex it is to paste things together. ‘Short reads’ sequence each chunk multiple times (10x or more for some Complete Genomics projects) while ‘long reads’ get away with doing that less. Long reads, however, also tend to be slower and more expensive. As Reid explains, long reads are good for some things, but Complete Genomics isn’t into those things. They have improved software that makes short reads the most economic way for them to proceed with human genomes.

This company knows exactly what it wants to do. And it is doing it really well.

That doesn’t mean I’m completely free of concerns. I’ll stop my blatant cheerleading for a second and do a little math. Reid didn’t give us an exact total for how many genomes they’ve sequenced (that’s ok, we know from other press materials that the number is more than 500 – damn impressive) but he did tell us that his current capacity is around 400 genomes per month. Those sell at a price of $10,000 each (or less if bought in bulk). Assuming there’s enough market demand (which I think there is), that means with current levels of production Complete Genomics could make around $4 million a month in revenue. So, maybe $48 million a year? Remember, even as Complete Genomics increases its production to drive up their revenue, the price per genome is planned to recede. This company has spent “tens of millions” on improving their software, and according to their own reports they spent more than $20 million in Q3 of 2010. Is $48 million per year enough for Complete Genomics to grow? Reid says they are very happy with their IPO, but I know that if I were in his place I would have been even happier if they had raised their original target of $80M instead of the $50M or so they got.

I think this next year is going to be a critical time for the company. Depending on funding and revenue they could have to stay on a conservative growth curve. Or, if they secure the finances, they could really ramp up – trust me, the technology looks like it is more than ready.

I really want a whole genome sequence to cost less than $1000 and not simply because I ‘m itching to peer at my own software code. By the time we hit that price point, researchers will have had the opportunity to look at thousands (tens of thousands?) of genomes and finally get the data they need to puzzle out some of the secrets of our DNA. That future may be nearer than we think. For now, Complete Genomics looks like the company that is going to take us there. I can’t wait.

Update: No person at Singularity Hub currently owns any stock in Complete Genomics or owns any financial stake in the company whatsoever.

[image credit: Complete Genomics]
[video credits: Aaron Saenz/Singularity Hub]
[source: Cliff Reid/Complete Genomics]

Open up your financial umbrellas, Complete Genomics is going to make it rain. The Mountain View startup has built a name for itself as one of the premier providers of whole genome sequencing for humans. Now we are just days away from their IPO. According to the filing statements with the SEC, Complete Genomics will offer 6 million shares of their stock at a price between $12 to $14. To encourage prospective investors to leap at their offer, the company released details of its current and future production. At the beginning of the year, the worldwide total number of human genomes ever sequenced was less than 300. Complete Genomics produced that many in the third quarter of 2010 alone. They hope to produce 400 genomes per month by the end of the year. These are big numbers, and they’re likely to get bigger. Much much bigger. In our interview with CEO Cliff Reid back in January, he claimed that Complete Genomics would sequence 1 million human genomes by 2014, and at prices substantially lower than any on the market today (possibly < $1000). The race to establish whole genome sequencing supremacy is underway, and this IPO will be a sign of how much faith the public has that Complete Genomics can come out on top.

It’s been a big season for genome sequencing. Ion Torrent, a company developing CMOS based DNA sequencing technology was purchased by Life Technologies for $375 million. Critical parts of the CMOS approach to DNA were actually licensed to Ion Torrent earlier in the year by DNA Electronics, a UK company looking to develop handheld genetic scanners that we’ve discussed before. Pacific Biosciences (NASDAQ: PACB), which has developed optics based DNA sequencing tech, had its IPO at the end of October and raised around $200 million. BGI, China’s premier genome institute recently announced it was teaming up with OpGen to further their own optical approach to DNA sequencing. Everywhere you look, from the EU to California to Asia, forces and finances are gathering to see who will provide the next generation of genome testing and analysis.

This heightened activity in the field may be just what Complete Genomics needs to fuel their IPO. With Ion Torrent selling for $375 M and Pacific Biosciences raising $200 M at a similar share price (~$16), Complete Genomics gets a good idea of what it can raise. The $86 million it hopes to pull in through its stock offering will nearly double what it has gained through venture investments. Complete Genomics gathered $39 million in venture funding this August putting it up to around $91 million in total. With another $86 M, Complete Genomics would have the funds to expand its new sequencing centers aggressively – a key requirement if they are to develop as quickly as Cliff Reid seems to be planning. Investors may look at Complete Genomics’s competitors recent financial gains, compare their tech to CG’s rapidly growing capabilities,and flock to the IPO.

Yet such investments are not without their concerns. Even while raising venture funds, Complete Genomics was fighting off patent infringement lawsuits from Illumina - perhaps their main rival in whole genome sequencing. Genetic testing based on SNP (single nucleotide polymorphisms) has faced growing concerns over accuracy and relevance in the light of real and perceived errors. The field of genetics, in general, has faced criticism for the lack of real world benefit for patients in the ten years after the first human genome was sequenced. Investors may see Complete Genomics’ cheap whole genome sequencing, which provides phenomenonally more data than SNP tests, as the technology that will come to dominate genetics and reconfirm its importance in medicine. Or they may see the uncertainty in genetics as a great reason to avoid investing in the field altogether.

The same factors which make investments a risk make them very exciting to techno-optimists like myself. I can’t say for certain that Complete Genomics will be the undisputed leader in whole genome sequencing. There’s too much potential competition from Illumina, and the CMOS and optical approaches developed by others are too attractive looking, to call things in Cliff Reid’s favor at the moment. Still, I do think Complete Genomics has the right approach to sequencing: specialize in one field (human genomes) and use economy of scale to push towards ever cheaper and larger production. 400 genomes a month by 2011 is an amazing accomplishment, especially as the costs for materials for each genome may be as low as $1800. The only way Complete Genomics is going to lose is if some other company can beat that. Either way, scientific research and personal genomics will have won. Cheap and fast whole genome sequencing will provide us with huge amounts of new genetic data that we can use to understand illnesses, and provide better healthcare. When your genome costs less than $1000 to sequence, millions all over the world will be encouraged to get themselves tested and claim an informed ownership of their own genetic information. Win or Fail, Complete Genomics’ IPO is another sign that the next DNA revolution is near.

[image credit: Complete Genomics]
[sources: Complete Genomics, SEC]

Venter, Collins grace the Time cover, July 3, 2000

“It is fair to say that the Human Genome Project has not yet directly affected the health care of most individuals.” – Francis Collins, April 2010, Nature.

What’s in a genome?  Ten years ago, the completion of the Human Genome Project promised to usher in a whole new era of heath care.  Revolutionary gene therapies would soon conquer everything from cancer and heart disease to diabetes and autoimmunity.  A roll-call of our genes would unlock the causes (and the solutions) to death and disease.  But a decade on, most of these hopes have failed to materialize, and most of our lives haven’t changed.  So where’s the revolution?

A recent retrospective in Nature includes some sobering reviews by such genetic gurus as Craig Venter and Francis Collins.  Sure, there have been some significant gains.  In vitro genetic screening has greatly reduced the risk of many common genetic diseases at the pre-birth stage.  Risk factors for a range of adult diseases (including cancer) are coming into focus, and a host of new drugs have been developed.  But as scientists expected to find common genetic determinants underlying common diseases, they quickly discovered that the genome was anything but straightforward.  Instead, the genes behind disease have been shown to be highly complex and individually variable, even for widespread disorders.  There isn’t a SNP for cancer.

The problem is that currently, the field of genomics is data-rich and application-poor.  Thanks to companies like Complete Genomics, there is a flood of new genetic data and even more on the way – but we still don’t know how it works.  So far, the primary focus of interest (and funding) has been the most easily quantifiable advances, such as sequencing speed and costs.  Accomplishments in this arena have been impressive, but a complementary push for clinical applications is needed to sort through all of this genomic data that we still don’t understand.

The fate of commercial genetics hangs in the balance.  Companies like deCODE and 23andMe were born on the hope that laypeople might be willing to pay for a glimpse at their own DNA.  The bankruptcy of deCODE and troubling rumors about 23andMe raise the question of whether personal genomics is an industry born premature.  So far, their products feed a curiosity niche, not a utilitarian one.  When a genome points to little more than SNP-based correlations, few people can justify spending their recession-hit income on what remains a biotech novelty.

As Collins, Venter and others have suggested, a health care revolution requires bridging the gap between genomic data and its clinical utility.  Any disappointments of the past decade point to the directions of the next.  We’re learning that so-called “junk DNA” isn’t really junk, but can regulate the expression of other, coding sequences of the genome.  Untangling the various networks of gene regulation will illuminate the pathways which result in a given phenotype, pathological or not.  The roles of epigenetic processes are also undoubtedly complicating factors which will need to be better understood.

Most approaches over the past decade have used SNP chip analysis to identify mutations associated with particular phenotypes.  This type of analysis only looks at small parts of the genome, and has largely failed to identify the genetic determinants of most diseases.  The SNP chip approach will be phased out as whole-genome scans become faster and more affordable (costs should drop below $1000 within the next three years).  Complete Genomics aims to sequence 1 million human genomes within the next five years, and that’s a lot of data to crunch. Venter is calling for two ways of making better sense of this flood of whole-genome scans: more detailed phenotype analyses, and the development of computational tools that can link them to their genetic counterparts.

It’s interesting to note the parallel between difficulties encountered in genomics and neuroscience. Recent years have seen an increasing shift in brain science from localization (areas of the brain that “do” things) towards neural-network approaches.  Just as we’ll unlikely find a single gene that causes cancer, we’re not going to find the “irony zone” of the brain anytime soon.  Reconceptualizing both genomics and the brain as complex, interactive networks remains a necessary step to significant advances in either field (e.g. a health care revolution or AI, respectively). And despite these setbacks, we can expect big things on the way.

Genetics has already revolutionized our health care in certain respects. Preimplantation genetic diagnosis (PGD) has already made huge progress towards eradicating genetic disease before birth, a significant but often overlooked accomplishment. But more lies ahead. Coming decades will see the creation of genetic therapies based around the specific molecular details of a given disorder. Diseases such as pediatric cancer are already the target of multi-year genomic research, and more diseases will benefit from genomic research as costs come down. And as the genetic underpinnings of disease come into focus, personal genetics will also undoubtedly enjoy a second life – regardless of whether today’s companies survive to see it.

We’ve scanned the genome; what remains to be seen is what we can do with it.

Here is a short summary video from Reuters that may be of interest:

The IBM device would read single strands of DNA as they passed through layers of a microchip.

Sequencing your genome is going to be such big business that everyone wants to get in on it, even if they aren’t ready. In a recent press release, IBM announced that it is working to create a microchip that will sequence DNA by running it through tiny ‘nanopores’. The DNA Transistor will be able to sequence the entire genome rapidly and for less than $1000. While a working prototype of the chip won’t be created for three more years, IBM thinks that the theory and computation behind the concept is sound. If ultimately successful, the computer giant would launch itself to the forefront of the genome sequencing field. For now though, the company is just pushing an idea, not a product. I love it when companies compete in a field, but IBM’s got years of hard work before it could be a genome sequencing competitor. A fact that makes their press release seem premature at best. Still, you should check out the admittedly cool PR video after the break.

The first human genome cost around three billion dollars to sequence. Today, Illumina is offering to do the same for $50,000 and Complete Genomics is looking to a $5000 price tag under certain conditions by the end of next year. But the big hurdle is $1000. At that point, sequencing a genome will become accessible to almost anyone, and could see wide spread adoption in health and medicine. Using genetic information, doctors could provide personalized health care that would target illnesses and choose treatments which best suit your body. Knowing more, we could live much longer.


IBM’s proposed device would work by threading a single strand of DNA through a 3 nanometer diameter pore that is drilled through a silicon chip using an electron beam. As the press release admits, that idea is being explored by several companies around the world (including Agilent, Sequenom, NabSys, and Oxford Nanopore Technologies). The IBM innovation is to slow the progress of the DNA through the nanopore by using electric fields as gates. By controlling the speed at which the DNA moves, IBM hopes to read the strand. That reading would be accomplished by measuring the variation in electric potentials across the pore, which is uniquely different depending on if the base is A, C, T, or G.

Strands of DNA suspended in a solvent would be encouraged to pass through the nanopore due to an electric bias. Electric fields in the pore would control strand movement and read the DNA.

When I first looked at the IBM video describing the DNA Transistor, I thought, “I’ve seen this before.” I interviewed DNA Electronics CEO Christopher Toumazou about the combination of CMOS technology and genetics last month. The SNP Dr is a hand held silicon chip based device that reads DNA to find interesting genetic sequences. Unlike IBM’s device, it’s already been made and tested.

But there are big differences between the two devices besides the fact that IBM’s is still on the drawing board. First, the SNP Dr looks for single nucleotide polymorphisms, the single gene variations that have important impact on physical traits, drug reactions, etc. The SNP Dr does not sequence your entire genome as the DNA Transistor plans to do. If it helps to draw comparisons to companies, the SNP Dr is like 23andMe and the DNA Transistor is more like Illumina or Complete Genomics. Secondly, the SNP Dr works by using enzymes (polymerases) to cause protons to be released as a strand of DNA is read, and the silicon chip measures that current. The DNA Transistor would only use electric fields to read the strand as it passes through the nanopore.

So the DNA Transistor could be an important improvement over the existing SNP Dr technology, but not yet. In fact, IBM is looking to have a working prototype in three years. That seems like forever. Complete Genomics will have $5000 sequencing by the end of 2010 and it should only get cheaper thereafter. The same goes for Illumina and every other genome focused company. By 2013, the $1000 hurdle could have been jumped by one of these companies. Heck, by that time, we could see the price drop much lower. If the DNA transistor is to be competitive that far down the road it will have to be cheaper (IBM has tossed around the $100 mark as a possibility) or much much faster.

Luckily for IBM, silicon technology has speed down to a fine science. I don’t know if the DNA Transistor is going to be the device that defines the next generation of DNA sequencing, but I do know that the marriage of CMOS and genetics is powerful. That’s a sentiment shared by the researchers behind the IBM device, Stas Polonsky and Gustavo Stolovitzky, in the companion video embedded below.

Everything about this press release strikes me as hype. Good hype, sincere hype, but hype all the same. There’s no doubt that the combination of genome sequencing and microchip technology is a game changing idea, but IBM doesn’t have a monopoly on it. Obviously DNA Electronics has a somewhat related device already nearing market, and there are other companies also working towards using nanopores to sequence the whole genome. A lot can happen in three years. Hopefully someone will have a working prototype by that time. It could be IBM, but no matter who it is, their cheap and quick sequencing is going to change our lives.