May 05, 2005
Helicos Biosciences May Offer $5000 DNA Sequencing By 2007
Cambridge Massachusetts 2003 venture capitali start-up Helicos Biosciences claims that by 2007 Helicos will be selling a machine that will sequence a person's genome for $5000.
Helicosís first commercial sequencing machines will be ready for sale by the end of 2006 or early 2007, says president and CEO Stan Lapidus.
Get your DNA sequenced in 3 days for $5000.
When Helicosís commercial machine is released, says Lapidus, it will sequence a whole genome start to finish in three days and for a cost of $5,000.
Currently the cost is in the tens of millions. If Helicos achieves their goal then prices for DNA sequencing will drop by over 4 orders of magnitude.
Helicos has licensed technology developed by then CalTech biophysicist Stephen Quake's group (Quake is now at Stanford Medical School). Quake's group published a 2003 paper in the Proceedings of the National Academy of Sciences that is probably what led to the founding of Helicos. Here is the abstract which describes a way to read individual bases (letters in the DNA alphabet) from a single strand of DNA.
The completion of the human genome draft has taken several years and is only the beginning of a period in which large amounts of DNA and RNA sequence information will be required from many individuals and species. Conventional sequencing technology has limitations in cost, speed, and sensitivity, with the result that the demand for sequence information far outstrips current capacity. There have been several proposals to address these issues by developing the ability to sequence single DNA molecules, but none have been experimentally demonstrated. Here we report the use of DNA polymerase to obtain sequence information from single DNA molecules by using fluorescence microscopy. We monitored repeated incorporation of fluorescently labeled nucleotides into individual DNA strands with single base resolution, allowing the determination of sequence fingerprints up to 5 bp in length. These experiments show that one can study the activity of DNA polymerase at the single molecule level with single base resolution and a high degree of parallelization, thus providing the foundation for a practical single molecule sequencing technology.
Solexa, which is mentioned in the original article above as planning to get to market with a DNA sequencing machine before Helicos, is also pursuing a technology for reading individual strands of DNA.
Helicos BioSciences, a tiny company in Cambridge, Mass., is beginning an ambitious effort to sequence single molecules of DNA by running them through microscopically small channels. (Other techniques generally require billions upon billions of copies of the target DNA.) So is Solexa, a U.K. company whose technique involves attaching stretches of a single DNA molecule to the surface of a chip and analyzing them via laser light and fluorescent tags that identify particular DNA "letter" sequences.
Incredibly cheap DNA sequencing will be occasion for a massive biomedical and social science project to compare the DNA sequences and large amounts of biomedical and behavioral and other information between millions of people. By comparing DNA sequences alongside large quantities to detailed information we will be able to find genetic variations for general and specialized intelligence, personality, disease risks, criminal records, career paths, other behavioral tendencies, esthetic preferences, and numerous other human characteristics.
The DNA sequence comparisons combined with detailed comparisons of other characteristics of individuals will produce results that will shatter the politically correct beliefs that now dominate academic social science. The political Left will suffer a greater undermining of their beliefs than the Right. But conservatives and libertarians will also find many of their beliefs challenged by science. Many religious beliefs about human nature will also be challenged by contradictory evidence found in our DNA. Humans will come out looking far more determined by nature. If Helicos achieves their 2007 goal then by 2010 the political debate in America, Europe, and in many other parts of the world will be irrevocably changed.
Thanks to Brock Cusick for the tip on Helicos.
I'll be really impressed if they meet that time line. Everyone will want to buy their machines. Have you seen any data on sequencing error rates?
No, haven't seen any error rate data. I don't expect them to release that kind of data at this stage.
Note that Quake's original work was a very early prototype for much smaller run lengths. So it doesn't tell us much.
I wonder how many times they think they have to sequence a person's DNA to get correct data. Keep in mind that any single cell used for a starter sample is going to probably have some damage to it that is unique to that cell. That is another significant error source. So multiple cells would need to be separately sequenced and cells would need to be chosen from locations that are least likely to have mutational damage.
As an aside, making complete sequences this cheap and rapid would allow for a better confirmation of DNA mutation accumulation levels in various tissues at various ages. An accurate determination of the background error level would be needed, but above that, we could check how many mutations cells have at various ages. For example, getting mutation levels in 1-, 5-, 10-, 15-, 20-, 25-, 30, 40-, ..., and 90-year-olds could allow us to see if mutations increase linearly, exponentially, or if the accumulation slows with age. Without apoptosis and senescence, we'd expect it to increase faster than linearly, but tumor supression methods may actually slow the accumulation at certain stages of life, even if the rate of mutation is constant or increasing.
Also, comparisons of mitotic and post-mitotic tissues would be interesting. And of course, one could posit that genes that are actively used in various tissues would have a higher propensity for mutation. We could quantify that with full DNA sequencing. (As far as I understand it, the mutation rates are implied from sampling and from tests that measure only one type of mutation, or types of pre-mutation lesions. Actually testing the entire sequence would let us track SNPs and other changes in the DNA sequence. I don't know about tracking non-nucleotide mutations though).
Knowing our entire genomic sequence or being able to compare our sequence with other people really doesn't give us much information. Proteomics needs to catch up before any of this information improves our lives to any great degree.
From the therapeutic standpoint, I'd have to agree. Designing a better drug won't be much easier just by knowing the base sequences of all the variants of given genes. However, from the diagnostic standpoint, we have a lot to learn from genetics alone, even if we don't understand how the genetics translates into proteins, enzymes, gene expression levels, etc., etc. Just knowing that someone has a higher probability for a certain disorder, or that they have a certain subtype of the disorder that wasn't previously known, will be quite valuable in medical diagnosis.
Aside from that, it will also help greatly in subsets of evolutionary research, including evolutionary theories of psychology and aging.
We have a lot to gain from this, even if the gains are not directly beneficial to therapeutics.
Jay, I agree that having the human genome sequenced has and will contribute to our understanding of human physiology. But, I don't think it's valuable for the average person to be able to sequence their ENTIRE genome. Simple genotyping for specific mutations or polymorphisms would be enough to categorize them into groups for more effective medical treatment.
Cheap DNA sequencing will very rapidly improve the quality therapeutic choices and change new drug development. One reason for this is obvious: A lot of drugs are kept off the market because they have very harmful effects on subsets of the population due to genetic variations that cause them to absorb or break down or concentrate them differently or due to some other cause. With cheap DNA sequencing we'll quickly find out which polymorphisms are responsible for toxicity. Then drug approvals will include instructions on which genetic variations are incompatible with a drug and which dosing to use for other variations.
Even existing drugs on the market will be used differently. Why do some people have better results with one drug or another? In some cases the cause will be found in polymorphisms.
But some (in fact the vast majority) of the polymorphisms that affect how well drugs work are not known. Some are going to be extremely specific to each disorder or drug. They won't be as easy to know where to find as obvious targets such as, say, cytochrome p450 alleles. With cheap full genome sequencing all genetic variations will become comparable. This will lead to the discovery of many more variations that matter for drug development and for therapeutic decision making.
What do you think of the possibility that with so many polymorphisms due to be discovered, we could actually end up over-stratifying and thus make it even more difficult to develop drugs?
I admit to being a pessimist when it comes to having hope that genomics will have any direct benefit for us in the near future. In my experience, everything moves at a glacial pace.
What exactly do you mean by over-stratifying?
I expect to see pharma companies develop drugs that, say, only 90% of the population can use due to genetic problems in the other 10%. I also expect to see drugs aimed at 50% and even 10% or 1% of the population. The smaller the percentage the drug is aimed at then the higher the cost is going to be for that percentage.
I do not see how such a development could leave us worse off. What happens now is that a drug is released, we discover a few years later that is it is dangerous for 1% of its users and we have no way of knowing who is in the 1% and who is in the 99%. So then the FDA pulls the drug off the market and we don't have the drug at all. Better that we can at that point discover what is genetically different about that 1% so that the other 99% can go on using the drug.
Glacial pace: You need to step back and look at technological history. Lots of innovations follow an s-shaped curve of uptake. If you'd been watching the computer market in the 1960s and early 1970s it would have seen like the pace was glacial in terms of bringing computing to the masses. Then chips finally became powerful enough to put a complete processor on a chip and prices fell low enough that the Apple II and IBM PC came along and things began to take off. Technology reached a critical mass.
Or look at cell phone use. Recently I read somewhere that only one out of over 200 people in the United States had one in the early 1990s. Now the people who don't have them are headed toward being the minority.
Or look at the internet and how fast it took off in just the last 10 years.
I expect DNA sequencing to go thru a very similar transition. Some time in the next 10 years more people in the United States will know their complete DNA sequence than do not know their sequence (or, rather, more will have it on a disk than do not have it). Prices for sequencing have dropped by a few orders of magnitude in the last 10 years. Sequencing costs look set to drop many more orders of magnitude in the next 10 years. At some point we will pass a threshold and the rate of use of sequencing technology will respond to the decreased prices by going from sequencing one mammalian size genome over a few years to sequencing a hundred thousand genomes per day.
The same processes that drove the computing and communications revolutions (advances in technologies making really small devices) are driving the geonomic revolution. Microfluidics and nanotechnology are making it possible to read a single strand of DNA on a microdevice and then to read hundreds and thousands of different strands in parallel. This will drive down costs by orders of magnitude for the same reason computing prices have been driven down by orders of magnitude.
Genomics seems useful to me only as a first step.
It seems to me that figuring out the epigenetics and the protein creation process is much more significant, especially considering the most therapeutic targets for drugs are based on proteins, not genes. I believe that a single gene can create up to 10 or more proteins from the particular gene. How does a particular gene result in making protein A in one situation and protein C in another? Developing a comprehensive "map" for this seems necessary for realizing personalized medicine.
"By comparing DNA sequences alongside large quantities to detailed information we will be able to find genetic variations for general and specialized intelligence, personality, disease risks, criminal records, career paths, other behavioral tendencies, esthetic preferences, and numerous other human characteristics."
I wonder what genetic mutation swept through the american population in the 20th century that turned all those farmers into factory workers?
The smaller the percentage the drug is aimed at then the higher the cost is going to be for that percentage.
This is exactly what I meant. Tailoring drugs for a small number of people isn't happening today. Hence, orphan drugs. Where are we going to find the resources to fund all the R&D needed and how will patients afford these more expensive drugs?
The examples you gave for fast technological advancement are interesting, but I don't think we can compare them to medical treatment. Any new medical technology or drug has the potential to cause debilitating illness or death. To guard against this, the testing and approval process will need to be long and rigorous. We can sequence like mad but that will probably slow down drug approval because of the clog in the pipeline from all the new drugs developed (assuming that actually happens).
Incredible ... this technology will dramatically change the manner in which a great deal of research is conducted ... It will enable people to modify the genes of other organisms very rapidly ... this creates the ability to take bioengineering to a scale never seen before ... Goingover your recent posts Randall a number of them will be impacted ... Coal/Oil conversion to methane is a great example
As things now stand drugs are developed that work for only a subset of their users. The drugs get approved and all users take them hoping that they are among those who will benefit. Some end up getting no benefit. Some suffer a net harm. Note that these people do not get benefit and even suffer harm right now without the sequencing information.
The difference with DNA sequencing is we will be able to predict who will benefit and who wll be harmed. The net effect will be less harm from side effects and less tim wasted trying treatments that will not help. Also, more drugs will be able to make it thru phases I, II, III to get to market because drugs that harm to only some subsets of the population while providing benefits to other subsets will not have to be cancelled.
If more drugs go into the pipeline then I do not see why that will slow down the rate of approval. The FDA can scale up. In fact, there is a fee mechanism for paying for faster approval that provides for more people to be hired to look at applications.
The problem with the drug pipeline today is that most of the drugs that enter phase I do not make it out of phase III with approval. The failure rate is well over 90%. I can not remember the exact figure but something like only 4% of the drug candidates that make it to clinical testing make it to market. Whether the failure rate is high due to safety or efficacy depends on the class of drugs. For cancer drugs (which represent about half of the pipeline today) most fail due to lack of efficacy.
Yes, I expect many radical changes from cheap DNA sequencing.
I have been meaning to mention one really important area which I've mentioned in the past: Mating practices will change. We will know a lot more about what we carry and what prospective mates carry genetically and will be able to be picky about each other in ways that are not possible to do now. Competition for choice DNA donors will become fierce.
Women will judge the DNA of prospective mates. Men will do the same with women. But women will have more choices due to sperm banks.
My guess is that most women will be unwilling to use sperm banks. But the percentage who do opt for sperm banks will be able to get choices far closer to their ideal. One reason is that women will be able to choose sperm donors who have two copies of many of the genetic variations they want. The women will not have to worry that a particular sperm won't have the genetic variants that make some guy smart. Just choose a guy who has two copies of each desired variant and it won't matter which of each pair of chromosomes end up in the sperm that gets used.
Cheap sequencing coupled with the ability to make lots of eggs, as Randall has written about here http://www.futurepundit.com/archives/002755.html#002755, will allow embryo selection. Consider an upper class or upper-middle class couple whose IQs average in the mid 130s. They are willing to pay for Ivy League educations should their children happen to get admitted. Would such a couple be willing to go through the expense of screening several embryos to find one or two who have no major genetic diseases and probable IQs greater than 145? Personally, if I had the money, I would do it. I think an extra 15 IQ points is a lot more valuable than a Harvard degree. There must be many people who think the way I do.
Yes, first the cheap sequencing will lead to the rapid identification of the significance of large numbers of genetic variations. Then the ability to test lots of embryos will lead to couples doing exactly as you describe.
The problem that the couple with the high IQs face is that their kids are more likely to be less smart than more smart than they are (regression toward the mean of a group's IQ is well established in the psychometric research literature). Give them a way to raise their kids' IQ and they'll jump at it.
Yes, a 15 point IQ boost is incredibly valuable and, yes, lots of people will quickly decide that it is valuable once there is a way to achieve it.
In many cases people tend to rationalize that making something better is not that important when they can not see a way they can change it. But give them a way to change it and they'll drop that rationalization really fast.
With embryo selection couples are going to have to trade off between looks, intelligence, personality characteristics, and health factors. They won't be able to get some combinations because the desired characteristics are on different members of chromosome pairs. Also, embryo selection won't allow you to choose from every possible theoretical combination. Each member of a couple has 2 to the 23rd power possible combinations they can donate and the couple as a whole therefore could make 2^46 possible combinations. But since some combinations would by chance come up more often the number of embryos that would need to be made to create all possible combinations will be far greater than that.
I think there is enough evidence today to begin to curb your pessimism in light of EGFR and Iressa. DNA sequencing is currently being utilized to get drugs targeted more effectively even though only 10-20% of Non-Small Cell lung cancer patients harbor this mutation. So small market sizes don't appear to be scaring off Pharmas. Look at the recent deal with Abbott and Astra Zeneca over Iressa. This demonstrates that Abbott is interested in diagnosing EGFR DNA mutations and Astra is interested in keeping Iressa alive even though the market share is theorectically small in patient number.
The list of mutations which predict drug response is growing like an S curve.
c-kit/PDGFR and Gleevac
bcr-abl/CML and Gleevac
Her2 amplification and Herceptin
just to name a few.
I agree with you on the epigenetics but keep in mind most of the sequencing technologies being developed will be the bets tools to study this. BiSulfite or methyl sensisitive endonuclease based sequencing strategies can all be deployed on the newer sequencing technologies.
Correlating drug response to genetics is the easy part. Utilizing it to infer mate selection via IQ...that's much further off.
I'm not a biologist, so it is entirely possible that I'm missing something. Once sequencing gets cheap, what is to stop researchers from gathering a racially mixed group of people whose IQs are greater than 145, a group whose IQs are between 95 and 105, and a group without physical symptoms but with IQs below 75; sequencing their genes; and then using computers to compare how the gene sequences of the three groups tend to differ? Wouldn't this point fairly directly to the genes that have major effects on intelligence? That isn't a rhetorical question. It is something I've been wondering about.
Yes, genotypes that are involved in determining IQ could be identified fairly quickly given a large enough group of people who have their DNA sequenced, their IQ tested using a few different test methods, and also their brains scanned to determine gray and white matter distributions.
See my post "Brain Gray Matter Size Correlated To Intelligence" to see why I'd want to include brain scans in such a study.
You'd want to include high and very low IQ people in such a test in proportions far greater than they exist in the general population.
I'd also like to see people who have a lot of humans included in such a study. I'm expecting that there are genes which influence inventiveness and curiosity which are not entirely the same as the genes that influence IQ. See my post "Low Latent Inhibition Plus High Intelligence Leads To High Creativity?" for another quality whose genetic roots would be useful to know.
Such a study really could begin now if the money was available. Tens of thousands of participants could be recruited and their tissue samples collected and frozen now. Their IQ and other cognitive qualities tested by a battery of tests, their brains scanned, their detailed medical histories collected, and assorted tests of beliefs and values preferences given. Also, detailed physical measurements could be taken including fine shade differences of hair, eye, and tissue colors, lengths of various bones and shapes of facial, ear, and other body features, fat distribution, and anything else measurable could be measured.
Also, instruments for automating the measurement of many physical features could be developed now. How about developing a machine that uses a safe type of laser that could scale the surface of a person to create a mathematical model of their surface shapes including lengths of arm and leg and toe and finger bones, head size, ear shapes, and everything else viewable externally. Such a machine could also measure skin color and distribution of intensity of skin color (which might detect anemia and skin diseases), freckles, skin hair concentrations, thicknesses, colors, and lengths, and still other externally measurable features. Could the device detect the difference between natural freckles and skin coloration differences caused by sun damage or physical injuries and burns? Probably.
Impressive, if it turns out to be true and precise. With this I'd imagine we'd not only be able to single out genes that significantly influence intelligence, but eventually in what way they do so(by analyzing the changes be it to a product protein or in regulation). This could lead us, guide us to further modifications(and novel combinations of genes) that could boost intellect far beyond that of any human.
My guess(some basis, not much) has been that at least an order of magnitude increase over the current max can be achieved this way.
Randall Parker is right on the money as far as to how some drug companies are envisioning their future (posts of 5/6 11:38am and 5/6 7:09pm). Last month, the FDA announced guidelines for submission of genomic data in clinical trials, clearing up one of the major uncertainties (e.g., providing some assurance that voluntary submissions of such data won't be held against the therapy being tested). The recent fates of the cox-2 inhibitors (Vioxx et al.) made people shudder, for the reasons R.P. described--effective pain relief for many, but unexpectedly increased risk of severe cardiac problems for a few.
At the moment, the genetic (or epigenetic) bases of such minority effects are mostly unknown, but the consensus is that the problem is now readily tractable with the tools at hand (e.g. genotype characterization by SNPs).
The (somewhat kitschy) terms "theranostics" and "companion diagnostics" are being used to describe circumstances where a diagnostic test is used to determine the suitability of a therapy for a particular potential patient. Some current examples were listed by Kevin McKernan (5/7 1:56pm). That list will be growing.
Nice thread indeed. I like Jay Fox's comment on clocking mutation rates. You could do this for different organs and finally have a metric on food or environments that are "bad" for you. DNA mutates all the time, but it is the bad mutation that kills you. The probability of the bad mutation has to be a function of the mutation rate, which, with cheap DNA sequencing, you can measure.
This company won't go far. They have no personality, no heart, and they're flat out lying about the capability of their product - and yet they've managed to get about 70M of VC money. What a shame.
Cheap? This will never be cheap, or fast. It's all still theory. At more than AN HOUR PER NUCLEOTIDE incorporation step (the truth shall set you free), how does this revolutionize anything? It all sounds so cool with the fluidics chamber and sparkly lasers and all, but to improve on current technology they need to get that incoporation per nucleotide step (flow in a labeled nucleotide, let it incorporate, wash out) down in the range of a second. It's not happening (by at least 2 orders of magnitude). Pipe dreams and pie in the sky. (sorry VC guys, big loss for you).