Congratulations on reaching the $1000 genome
August 3rd, 2012
Recently there has been a spate of talks, press releases, and articles about the absurdity of the $1000 human genome, e. g., Cancer, Data and the Fallacy of the $1000 Genome. No doubt this has contributed to the somewhat muted response to Life Technologies announcement that they will attempt to win the using the Ion Torrent Proton platform. While I agree that talk of the $1000 human genome is irrelevant, it’s not for the same reason as everyone else. Most people cite sequence analysis costs, not typically included in the $1000 per genome estimate, as the reason that talk of a consumables-only $1000 genome is not relevant. That is a red herring (but more on that later). The real reason that the $1000 human genome is no longer interesting is because, for all intents and purposes, we have already achieved the $1000 human genome. “What?!?” you say, “a human genome costs $5000 to sequence!” Sure, you’re right, but that is just details. Compared to $1 billion (the approximate cost of the first human genome), the difference between $1000 and $5000 is rounding error. The reality is that the current cost of sequencing a human genome is well within the cost of diagnostic tests in common use in health care. From another perspective, the cost of sequencing a human genome has fallen into the range of an expensive vacation, i. e., there are people who at present are getting their genomes sequencing for recreational purposes. So, congratulations, we did it!
Now, what of all this talk of $1 million or $100,000 to analyze the sequence data from a human genome? Does analyzing sequence data from a single human genome cost that much? Well, it certainly can, but it need not. And particularly for the clinical market, it certainly won’t and it is preposterous to posit that it will. The confusion arises from people failing to distinguish between research and clinical analysis. While research projects on cancer are trying to better understand cancer, to expand our knowledge of the disease, the clinical application of genome sequencing to cancer, or any other disease for that matter, will be focused on improving diagnosis and treatment. The current reality is that we know precious little about how the genome works; and the ability to translate this information into improved diagnosis or connecting that diagnosis to a treatment is even less. In other words, the amount of actionable information one gets from the genome or transcriptome sequence is relatively small compared to the massive amount of presently uninterpretable information. Cancer research and more fundamental investigations into how the genome works (e. g., ENCODE) will expand our knowledge and potential actions over time, but for now there are only dozens to hundreds of possible actionable outcomes. Bottom line, the amount of analysis to convert genome sequence into these possible actions should not cost more than $100. So if you are a real stickler on getting the total cost of sequencing and analysis of a clinical human genome below $1000 (or less than one-third what Myriad charges to assay a few genes in their BRCA test), we just need to get the cost of sequencing less than $900 and we’re there.
This all is not to say that the cost of sequencing and analyzing a human genome does not matter; it most certainly does. In research, reducing the cost by a factor of two means you can double the number of samples you are able to sequence; potentially greatly increasing the power of your study. So we must continue to strive to reduce costs, but like I said, for all intents and purposes we have already hit an amazing milestone. So enjoy your weekend, you deserve it.
Posted in genomics | 6 Comments »
Tagged with: genomics, health, informatics, science
You can follow any responses to this entry through the feed. You can leave a response, or trackback from your own site.
August 4th, 2012 at 10:24 am
David, welcome back!! Guess I need a new meme…
Speaking of rounding errors, the catchy phrase “the $1 million interpretation” — coined by former ACMG president Bruce Korf in my book — wasn’t intended (a) to be taken literally, or (b) to refer solely to the computational
cost of a clinical genome interpretation. It’s more about building the broader medical infrastructure, ensuring the quality of genome-wide information and developing the resources to communicate those results to the patient — on the assumption that this is heading to widespread (universal) adoption.
More specific concerns include confirming the significance of variants of unknown significance (including functional assays); urgently growing our meager genetic counseling resources; and training a new generation of physicians about what this all means.
In the book, Korf said: “Let’s imagine you could sequence an entire genome for a modest cost, now here you are ready to counsel them. The problem is, for the most part, you wouldn’t have any good data to base counseling on. You’d have a fantastic amount of information, but not a whole lot of background about what it all means.”
He continued: “No-one’s going to be able to read the sequence and say, ‘This is what you’re going to get, this is what you’re not going to get.’ It’s going to take at least a generation, possibly more, before we have a clear handle on that. We’re deciphering an extremely nuanced, complicated code.”
Personally I doubt that amounts to $1 million/genome, but it’s not trivial — or irrelevant. In the meantime, cheers!
Kevin Davies
Author ‘The $1,000 Genome’
August 4th, 2012 at 9:27 pm
Kevin,
Exactly right. The intent here is not to cast aspersions on the $1000 genome meme, but to point out that people are confusing clinical and research genomic analysis. You and others clearly make this distinction, but many don’t. There is much that needs to be done, and it will be costly, but it will be done in research labs, not clinical labs.
August 5th, 2012 at 11:09 am
Decades ago the goal of a $1,000 personal computer was achieved by meeting that price point, not by landing within the margin of error of the cost of building the first computer (whatever that may be).
August 6th, 2012 at 9:02 am
Very interesting post, and it links up nicely with Daniel MacArthur’s post from a few months ago at Genomes Unzipped: http://www.genomesunzipped.org/2012/05/genome-interpretation-costs-will-not-spiral-out-of-control.php There’s some excellent conversation in that post and in the comments about what it really means to interpret a genome, which I won’t repeat here.
I do, however, want to make a comment or two about the distinction you draw between clinical and research uses of whole genome sequencing, which I think is extremely important, both because of that is entailed and the implications for the “$1,000 genome/$1M interpretation” meme.
As you say, WGS for research brings with it a desire for a much more generalized interpretation – which can drive up the nominal cost of such work – but it also carries, arguably, a more acute price sensitivity since studies are funded for dozens or hundreds of genomes and cutting the cost in half, even if that’s just a “rounding error,” makes a big difference.
Clinical whole genomes may well carry a more limited interpretation requirement. As you say, “The current reality is that we know precious little about how the genome works; and the ability to translate this information into improved diagnosis or connecting that diagnosis to a treatment is even less.”
But the ability (or inability) to interpret WGS data at an individual level – and thus the extent to which the genome can actually be used as a diagnostic and/or to drive therapeutic actions for a particular individual – is directly correlated with the reimbursable value of the genome.
Remember, on the clinical side, unlike on the research side, (in most cases) genome sequencing and interpretation ultimately needs to turn a profit, or at least be able to break even. Even if it’s true that “…the amount of analysis to convert genome sequence into these possible [clinical] actions should not cost more than $100,” what matters, ultimately, is whether payers will be willing to reimburse for that analysis at a rate that allows the provider of the WGS service to turn a profit.
And obviously, here, we are talking about more than profitability in the sense of recouping the sequencing and analysis cost for a single genome. You also need to consider the amortized cost of all of the work done to develop the WGS product, not just scientific/technical but also infrastructure costs (as you know and have written about before), regulatory compliance costs (e.g., CLIA certification and potentially FDA clearance), reimbursement costs (including convincing payers that there’s any value) and all of the other transactional costs associated with bringing a product to market.
I’m certainly not attempting to suggest that the clinical interpretation of a genome does – or should – cost $1M or $100,000 or any other number. I’m just trying to point out that, particularly on the clinical side, what really matters is the relationship between the cost of clinical analysis and the value that can be demonstrated to payers.
August 6th, 2012 at 10:10 am
Theressia McClure, that is true, but it is also true that $1000 then bought significantly less than $1000 now. A more appropriate measure is that the current cost is within what one expects to pay for clinical diagnostics and luxury items.
August 6th, 2012 at 10:15 am
Daniel, yes, showing efficacy/value is important, but this has already been established for some targeted sequencing tests. So the remaining hurdle is to show that the additional cost for whole genome sequencing garners sufficient additional benefit. This will likely first occur in cancer, where focal copy number and structural variation can play a significant role.
All the other sources of cost you mention certainly need to be considered, but at the scale of clinical application, they should be manageable.