Personal Genomics, tipping points and a personal perspective Please indulge a long post from a personal perspective, what genomics is about to do for _me_. This is information that many, if not all, of our readers already know. I’ve been researching and working in either experimental biology or genomics for over 20 years. Ever since the beginning of the Human Genome Project , which coincidently started the same year I started my Ph.D. program, into my postdoctoral research at EMBL and now my work at OpenHelix, I’ve known that someday personal genomics was going to impact me, and millions of others, in a big way. Yet, it has always felt that it was one of those things that would be a decision I and we as a society didn’t have to make until we turned that corner that seemed always “just ahead.”

But now I think we’ve turned a corner. It feels, to mix metaphors, that we’ve hit a tipping point. The Human genome project, the mapping and sequencing of the/a human genome from 1990 to 2003, cost approximately 2,700,000,000 dollars (that’s 2.7 billion, I wanted to get all the zeros in). Celera did the genome for 300,000,000. The cost of sequencing an entire human genome has been plummeting ever since. In 2007, the cost of sequencing the genome of James Watson (co-discoverer of DNA) was about 2,000,000. The today cost is about 10,000. Complete Genomics and other companies are on the march to quickly reducing the cost of sequencing a genome under 1,000.

Let me graph the last 8 years for you, mind you, this is starting from the 300,000,000 number, not the 2.7billion, because that graph would be a straight line down.

So, within a year, the cost of sequencing your, my, genome will reach 1,000. If not less. We’ve seen this coming for years now, and it’s upon us. But what does it mean?  A lot of data. But data means nothing without context and analysis. Sequencing my genome would be a waste of 1,000 dollars if I gleaned nothing from it.

Yet, even that seems to have turned the corner from a few tidbits of genetic information to a steady steam and the beginning of a flood.

You know you’ve turned a corner when a genomics testing company begins to offer genetic tests to the mass market through Walgreens. There’s enough context in that data to make money from it, or so they hope. You can be sure the corner is safely behind you when the FDA tells Pathway Genomics and Walgreens that they will need to hold off while they make sense of the regulatory implications. Genomic ancestry test are are also gaining is usability… and scrutiny.

It was the recent Lancet paper on the clinical analysis that seemed to be a tipping point, not for me or those in the field. Genomics has been on my radar since 1988, but for society. I blogged about the paper and it’s use of genomics resources such as GVS, dbSNP and others. In the paper, the researchers did a thorough clinical assessment of an individual’s genome. We’ve brought down the cost of sequencing, now we are learning how much it’s going to take to assess that data from a medical point of view, and importantly, what we can learn from it.

What can we learn from it? I read this paper again from a personal perspective now. Could I learn something from sequencing and analyzing my genome, and if so what. My answer came to this: yes, I could learn something and in fact enough that I’m not convinced that as soon as that sequencing gets down to a 1,000 or lower (and is a high quality sequence :), I’m going to do it.

There are three things I see from this paper that one could learn from assessing their genome: prevention, early detection and therapy. I believe the former will be, for most people, something they already know and their genome sequence will tell them nothing new. The other two could be a wealth of information they will want, even need, to know. You’ll notice I left off ‘cure.’ I saw nothing in this paper, and nothing on the near horizon, that suggests to me that our genome sequence data will help with curing anything. Perhaps, just not much. Yet, the possibilities of early detection of disease and personalized drug treatment are tantalizing.

Prevention: The authors have a graph. In the middle of the circle are all the diseases that this individual has a high propensity or probability of contracting, the size of the text indicates the increased probability of the individual getting the disease. For this individual, the biggies are type 2 diabetes, obesity, osteoarthritis and coronary heart disease. Of course, these and others also inter-relate, obesity adversely affects hypertension which adversely affects coronary heart disease, etc. I think my chart might be somewhat different. I know hypertension would be a large one. I’ve had hypertension since I was a skinny 18 year old and the doctor measured my blood pressure at 200/160. That’s going to be a big one. From my family history, I’m sure coronary heart disease will be a large one having had several people in my family have heart attacks at early ages. I am not sure what else will big for me in that middle circle.

Around the circumference of the circle are environmental and lifestyle factors that could have an effect (positive or negative) on the probability of the person getting the disease. A guideline of sorts for the things this person needs to do to lower his chance of getting these diseases. Perhaps it’s a function of the specific diseases he has a propensity towards and other’s might have different ones that would have different lifestyle and environmental factors involved, but I’m think it’s going to be a safe bet that it won’t be much different for the majority of people:

Those are his four biggies of lifestyle and environmental changes he could make to lower his chance of getting these diseases. We’ve known this for decades, if not most of human history. Diet is simple (if not easy), lots of whole grains, fruits and vegetables, nuts and legumes garnished with a bit of dairy, eggs and lean meats. Limit the processed foods piled with fat, salt and sugar. Exercise is simple (if not easy), a moderate daily activity. Stop smoking, calm down.  Nothing new here, move along.
As I’ve struggled with hypertension and weight over the years, I’ve learned this. It’s pretty simple and straight forward and our doctors have been telling us this for years. If you smoke, stop it. Change your diet, start exercising, calm down. It’s simple. It’s not easy. It’s been a Herculean task for me over the last decade, but I’ve lost 50lbs, my diet has changed almost exclusively to what I mention above, I’ve incorporated walking several miles a day into my routine and I took up knitting (to lower stress, with the added benefit of making things).

My take home from this is that sequencing your genome will tell you nothing you probably don’t already know. So what will it tell you that you don’t know and that could help you?

Early Detection: As with the individual studied in the paper, my family history tells me a lot. Sequencing my genome in many ways will tell me little I don’t already know. I have a family history of hypertension, heart disease, prostate cancer and a couple other things. But don’t I know. Interestingly, from the paper they discovered from this man’s genome that he had a much higher propensity to contract hemochromotosis. Quoting:

Analysis of the patient’s genome revealed three novel and potentially damaging variants in two related genes that were previously associated with development of haemochromatosis. Subsequent to these findings, detailed review of personal and family history did not identify a history of haemochromatosis in the patient or family members. Echocardiogram results and liver function tests did not show evidence of the disease.

In other words, he has a heightened risk of a disease that could kill him. He could have never known that from his family history and had no indications of that from medical tests. Left undiagnosed and untreated it could lead to heart failure, liver disease and more. The good news is that the earlier the condition is detected, the better the prognosis for treatment. This patient can now keep an eye on this through regular testing and treat any early symptoms.

So, I wonder what hidden probabilities I have, which tests would help detect an onset of an otherwise potentially life-threatening or debilitating disease. Perhaps knowing that alone would be worth the 1,000 dollars and save thousands more in treatment and testing. Of course there is the flip side, finding a propensity for a disease for which there is no cure, treatment or perhaps even test. For me, the benefit outweighs the cost. I find it relatively easy to ignore elephants in rooms, so I think I’d be pretty good at ignoring a 60% chance of an untreatable disease elephant lurking in my genome :).

There is another aspect to this for me. We have an adopted daughter. We know little of her family health history. Recently an entire family, two kids and their parents, had their genomes sequenced. Something we learned from that paper, and many of the others, is that we still have a long way to go. There are so many diseases for which we don’t know the genetic basis of, and so many genetic variations that we have no idea how they affect our health if at all. It’s a cautionary tale in the rush. Still, what information could our daughter learn from her genome that she couldn’t get any other way?

As Daniel MacArthur of Genetic Future (a must-read blog btw if interested in genomics) states, we have to do a lot more genomic sequencing before we have a good handle on the genetic cause of most common diseases. But at 1,000 dollars a genome, isn’t that corner just about in front of us? I’ll bet by 2015 we’ll look at the 1,000 genomes project as a miniscule number. By the, I’ll make the prediction right now… we’ll have 10,000 genomes and the data we can glean from that.

Treatments and Drugs: So, not much with helping me with prevention, some good news possibly with detecting disease now and definitely in my near future. The other, and possibly one of the most promising, aspects of my genome test is picking the drug and treatment for the right person. Let me relate a personal story regarding hypertension. I have it. I can alleviate it by eating right, exercising and reducing stress. But it will still be high (perhaps 150/110 instead of 200/160), and I’ve been working on that. Didn’t need a genome sequence to tell me that. I have a family history of it, didn’t need to have a genome sequence to tell me I would have a good chance of having it. But even with detection and lifestyle changes I need drugs.

When I moved from Germany to the US 7 years ago, my new American HMO would not approve a drug I had been taking and instead I was prescribed a calcium blocker. Within a few short weeks, my right leg swelled up to twice it’s size, turned purple, dry and painful. I was taken off the drug, and though my leg got back to normal size within a year, to this day my foot and knee are discolored and dry.

Drugs affect differently people, differently. Doctors know that. For the most part, it’s experimentation. My doctor gives me one set of hypertension medication, adds another, changes the dose, removes one, adds another… until finally he gets a combination and dosage that works for me… hopefully without any annoying… or threatening… side effects.

This paper points to the promise of being able to tailor the drug treatment for a patient based on their individual genomic character. This patient’s genome suggests that if he ever needed Warfarin, he’d need a reduced dose. It also suggests that if he ever contracts Type 2 diabetes (for which his genome suggests he has a high propensity for), Metformin and Troglitazone probably will have a reduced likelihood of response.

I doubt my particular side effect caused by the calcium blocker would have been detected in my genome, and the science here too is still in it’s beginning stages, but the promise is enough and the amount of data we are getting is growing exponentially. Perhaps by the time my genome is sequenced, and I contract some condition that needs treating, I’ll be able to have the drugs tailored specifically to me.

Tipping Point: In the last couple months we’ve seen clinical assessments of individual genomes, an entire family’s genome sequenced, and several other papers on individual genome sequences. We’ve seen Pathway Genomics offer genetic tests at Walgreens and the FDA balk. GWAS studies and individual genomes and 1,000 genome projects have been coming at a dizzying pace. Though it’s been coming for 20 years, and those of us in the genomics field have been watching it and studying it, and in turns been frustrated by the pace while at the same time looking back and awed by it, so it’s nothing new to us. But I believe that we will be looking back at 2010 as the year that personal genomics tipped into the mainstream conscience. I’m also convinced that the mainstream, the medical profession, the patients, won’t be ready for it. We’ll and our doctors will need a ‘genome browser’ to look through our data an. Interestingly, the UCSC Genome Browser has just included the personal (but public) genome variations of 9 individuals. I’m kind of geekily salivating at the ability to do that with my own genome :D.

On one last note, I have to say that GINA and Health Care Reform came none too soon. When we have tens of thousands of genomes sequenced and patients and doctors combing them for data and information, genomic privacy will be of paramount importance. And when risk is now known instead of unknown, and _everyone_ has a pre-existing condition by virtue of the genome knowledge, health insurance reform was necessary to spread the risk (and keep the system from breaking). Neither are enough, there is more to be done, but a good foundation to ready us for the coming.. the arrived.. age of personal genomics.

Ashley, E., Butte, A., Wheeler, M., Chen, R., Klein, T., Dewey, F., Dudley, J., Ormond, K., Pavlovic, A., & Morgan, A. (2010). Clinical assessment incorporating a personal genome The Lancet, 375 (9725), 1525-1535 DOI: 10.1016/S0140-6736(10)60452-7

17 thoughts on “Personal Genomics, tipping points and a personal perspective

  1. Steven Murphy MD

    The best thing you did for yourself was lose that weight! I have to tell you, I am a little frustrated with the research community here. Everyone keeps saying how fantastic this paper is.

    It is excellent that they showed “how” to analyze a genome given today’s tools. But clinically it is not very good.

    1. Echo and LFTs are not how you evaluate for Hemochromatosis damage.
    You check Iron studies, you may need t o do a liver biopsy or MRI, you check things like HGBA1c and maybe even testosterone.

    2. It is no surprise that there is no family history of Hemochromatosis HFE or Non HFE. As you know, HFE disease is AUTOSOMAL RECESSIVE! Thus I expect no family history or a horizontal affected line.

    3. Even those with HFE mutations only manifest disease 20-30% of the time, thus even “Armed” with that genomic knowledge, it is not that clinically useful.

    4. Simple ancestry of Stephen may prove more useful. This was something completely LEFT OUT of his pedigree. If he is on northern european ancestry, I would be considering Iron studies to evaluate for Hemochromatosis.

    5. Yes, god forbid Stephen have to be on Warfarin. It is nice to know his status. I agree

    6. The family history of SCD would alone be enough to prompt an echo, the genome scan adds nothing to that evaluation other than further prompt what should have been done…..

    In the end for me, this paper showed the proper way to clinically evaluate the genome and the improper way to clinically evaluate a patient…….


  2. Trey Post author

    first.. a clarification :D, as I commented over at Gene Expression, I was indulging in a bit of optimistic hyperbole with the 1,000 in 1 year prediction… though I’d don’t think it’s too optimistic to say 2-3 years. And obviously, as also commented by another, $1,000 dollars to sequence might be possible, but the cost of analysis and interpretation of that data will still be very high (heck, the number of people on this one analysis!), though I’m sure that will come down as new tools are created.

    @Steve. Thanks for the comments. I have a question about the hemochromatosis since I’m not a doctor. They weren’t necessary checking for liver damage per se, they were checking for indications of Hemochromatosis, pre-symptom, were they not? Wouldn’t the tests called for be different? The Mayo Clinic page on this says a simple serum transferin saturation test will detect before damage occurs. I’m not sure what they mean by “liver function test” in the paper, is that part of such a test?

    As to family history, recessive traits, even ones with low penetrance might or might not show up in a family history. But isn’t that the point? They often don’t and a understanding of an individual’s genetic makeup might indicate things that aren’t showing up in family histories. And for those of us without family histories, even that much more important.
    And hereditary conditions are often caused of course, even more often than not, by things more complicated than one autosomal recessive allele of a single gene. This patient had several variations in two genes that affect hemochromatosis.

    Do doctors usually suggest tests that would indicate hemochromatosis regularly? (honest question. I’m a 50 year old who goes to my doctor semi-annually and has blood tests done as a matter of course, but have no idea if anything tested for would give a good indication of hemochromatosis)

    I agree with you, they do not indicate whether they looked at his ethnic background (though perhaps it was part of just looking at his family history) and what that might indicate. That would seem a no-brainer. It’ points out one of the things that bugs me most about these papers (almost all research using genomic resources). They rarely, if ever, give a protocol or history of exactly what was done. There is little reproducibility. That’s something that really really needs to change.

    I honestly don’t think the paper told magnitudes more than would have been possible from a complete and full work up, but I do think that it (along with the myriad other papers, news and research coming out recently) does give a good picture of what genomics will be able to do for us. It can’t do it now because a) the cost of analysis is still going to be astronomical on a per person basis and b) there’s not enough data (association studies, drug studies, etc). But as the cost plummets, and 1,000 genome projects are completed and more and more people do it… both a and b will be taken care of and the things indicated in this paper will happen.

    It’s a tipping point this year. Not a fulfillment of the promise, but does feel like a tipping point to it.

  3. Pingback: Friday SNPpets | The OpenHelix Blog

  4. Pingback: Daddy, Papa and Me » Ancestors

  5. gsgs

    how do I analyse and compare human genomes of 3GB ?
    how many nucleotides are typically different
    between 2 average humans, 2 average siblings

  6. Trey Post author

    @gsgs, first comment: I have no idea what the average difference between siblings would be, perhaps you could google that. 99.9% of the human genome is identical between human individuals, so that would give you approximately 3,000,000 possible nucleotide differences. Though difficult, it’s obviously possible of course to compare and analyze. To see how to compare and analyze, this paper cited is a great place to start. My earlier blog post goes through some of the tools they used, and the UCSC Genome Browser now has individual genomes in it as tracks (and the Neanderthal) for comparison.

    @gsgs, second comment: no :-D. The graph is just there for impact, no real analysis going on. If you want the specific numbers and years, I can give those to you and let you put them into a logarithmic scale :-).

  7. Pingback: Regulating DNA (tests) | The OpenHelix Blog

  8. gsgs

    so we can store a genome with just a few MB
    by giving the differences to some referrence-genome ?
    e.g. to the first human genome at genbank
    I’d like to download some of these difference files.
    How many differences had the Nendertal genome ?
    Some (famous?) siblings should already have their
    differences been counted

  9. Trey Post author

    yes, basically, for the most part to the question (though some issues).

    Human/Neandertal is about 99.7%.

    If you look at the annotation tracks at the genome browser linked above you will see 2 siblings. One could catalog those differences using UCSC or Galaxy or the like.

  10. gsgs

    thanks, Trey.
    are these differences (like 99.7%) usually considered enough or do we need
    the exact positions to determine the relationship ?
    or nucleotide-differences and amino-acid differences.
    not the whole genome may be needed to estimate this %-difference, it should be usually the same % for
    the chromosomes (?)
    how many such differences are available ?
    how many complete human genomes are available ?
    checking the webpages and papers is a bit tedious …
    but I’ll try again later

  11. Pingback: Bulk sequence pricing | The OpenHelix Blog

  12. Pingback: Entering the personal genomics era | The OpenHelix Blog

  13. Pingback: 23andMe Scan: what I plan to find out | The OpenHelix Blog

  14. Pingback: What we expect from 23and Me genomic scan - Genomes Are Us

  15. Pingback: The cost of personal genomics - Genomes Are Us

  16. Pingback: Let the personal genomics bubble begin… - Genomes Are Us

Comments are closed.