This April, the field of molecular biology will celebrate two remarkable anniversaries: sixty years since James Watson and Francis Crick introduced the world to the structure of DNA, and ten years since the completion of the Human Genome Project (HGP). The former taught us what genetic material looks like; the latter taught us what it says.
Today, the HGP seems almost quaint—what took 15 years and nearly $3 billion to complete in 2001 can now be accomplished in about a day for less than $10,000, and soon that cost will likely drop below $1000. (Indeed, that line may already be crossed, if only the sequencing process itself is considered.) The NHGRI Genome Sequencing Program estimates its per-genome cost is $7666. Illumina, whose HiSeq DNA sequencing systems produce the bulk of human genome sequence reads, offers its sequencing services for as little as $4000 in bulk, while Complete Genomics offers its genomes for $5000 or less.
“2013 is likely to be the year where we see the $1000 genome. It's going to become that cheap to sequence the human genome,” predicted Daniel Franklin, executive editor of The Economist (and nephew of DNA pioneer Rosalind Franklin), in an online video (1) accompanying the magazine's “The World in 2013” issue. If nothing else, the barrier could fall this September when the $10 million Archon X Prize competition takes place. To claim the prize, a team must produce 100 human genome sequences in 30 days or less, at a maximum cost of $1000 per genome sequence (with an error rate of at least 10-6 and at least 98% completeness).
The concept of a $1000 genome was first articulated at the start of the millennium. It's a price point that a sequenced genome on par with a laptop computer or a clinical test—a psychologically pleasing, if not completely arbitrary, value. But just what does it mean?
“That is the first and best question on this topic,” says Chad Nusbaum, co-director of the Genome Sequencing and Analysis Program at the Broad Institute in Cambridge, MA. The term itself is nebulous: Is $1000 a cost or a price? Does it include only the cost of reagents or does it fold in sample preparation and labor? And what about the bioinformatics analyses, the genome interpretation, and long-term data storage?
The time has come to ask just what that $1000 will buy.
The $5000 genome
As a general rule, research and development (R&D) is most expensive at the start, when producing that first success comes at the cost of many initial failures. Subsequent attempts cost far less as failure rates fall, technology prices drop, and economies of scale kick in.
Genomics is no different. And if there is one sequencing facility that can count on economies of scale, it is the BGI in Shenzhen, China. BGI houses 128 HiSeq 2000 instruments, each capable of producing 600 gigabases (Gb) of sequence data per run. As the human genome is 3 billion base pairs long, achieving an average 30x coverage means researchers must collect around 90 Gb of sequence per genome. Thus, during each 10-day HiSeq 2000 run, scientists can, in theory, interrogate up to six human genomes, depending on coverage.
BGI declines to quote exact prices, but citing an online quote from Illumina for a New York–based medical school, Deputy Director Xun Xu estimates a single HiSeq 2000 run costs about $12,000 in consumables; assuming five samples per run, reagent costs run about $2400 per genome. That represents the bulk of genome sequencing cost today, Xu says; sample preparation and library construction are relatively inexpensive and largely automated, working out to $500 —1,000 per genome. In total, that adds up to around $3000 for reagents and labor, plus another $1000 or so for instrument depreciation.
Noting the apparent differences in genome pricing at different sequencing centers, Nusbaum —who also declines to quote a per-genome cost for the Broad Institute—says it is important to consider the “all-in” or “true” cost of sequencing: that is, not just the reagents, but factoring in everything from blood tubes to labor to instrument amortization. “You want to be very clear about how you are determining the cost,” Nusbaum explains. “There is a difference between the cost of doing the experiment, versus the true cost…for genomes, I think it's important to do the ‘true cost’.”