Cancer cells pull many dirty tricks to resist therapy and stump researchers. One is a madcap reshuffling of the 3 billion letters of DNA that make up the genome found in healthy human cells. But now that genome sequencing has become so cheap and fast, Mountain View, CA-based Complete Genomics is trying to enter a new market that would have been impossibly expensive a couple years ago: It is now seeking to help scientists pinpoint what’s different in the complete genome of a tumor, compared to a healthy cell.
Complete Genomics (NASDAQ: GNOM)—the sequencing company backed in its early days by San Diego’s Enterprise Partners Venture Capital and OVP Venture Partners in Kirkland, WA—is announcing today it is amping up its efforts to sequence entire human genomes as a service to cancer researchers. In addition to its sequencing service, the company plans to run basic computational analysis for all the copy number variations, and structural variations that pop up in tumors gone wild. This is essentially an extra computational feature that Complete Genomics is throwing in, at no extra cost, to its basic $10,000-per-genome sequencing service.
The idea in the beginning, CEO Cliff Reid says, is to save researchers the time and hassle of performing their own sequencing runs, and then looking around for proprietary or open-source software to help interpret that mother lode of data.
“This gives researchers a new avenue for understanding the genetic causes of cancer,” Reid says.
This strategic push into cancer is the first peep of news out of Complete Genomics since the company nailed down its initial public offering last month. Complete raised $54 million in that deal to move forward with its quest to become one of the leaders in the fiercely competitive field of super-cheap, super-fast DNA sequencing. The current market leaders—San Diego-based Illumina (NASDAQ: ILMN) and Carlsbad, CA-based Life Technologies (NASDAQ: LIFE)—have been built on business models in which they sell sophisticated instruments to researchers, plus chemical reagents to operate them. A highly-touted new entrant, Menlo Park, CA-based Pacific Biosciences (NASDAQ: PACB) is following a similarly proven business model, with a different technology that seeks to raise the bar on speed, and lower it even further on cost.
Complete is taking a clearly different approach in its business model. Instead of following the razor/razor blade model, it hopes to achieve super low cost and high efficiency by asking researchers to ship samples to its centralized lab where the sequencing and basic computing work gets done. This new business model for sequencing “is a change the industry is still digesting,” Reid says.
Part of what Complete is betting on here is that biologists will essentially want to farm out the heavy computational lifting that it takes to detect signals in datasets with 3 billion data points. Before Complete started offering this service commercially over the summer, the only alternative for researchers who wanted to look at variations in tumor genomes was to do the sequencing runs on an instrument, and either write their own software, or look around for a proprietary or open-source software program to help separate the signal from the noise, Reid says.
This is especially hard to do with cancer, Reid says, offering some basic science to help explain. Most healthy cells have two functioning copies of the genome. Sometimes they get deleted or possibly amplified five, 10, or 20 times over in tumors, through what is known as copy number variation. As more and more researchers have identified major structural changes in the genomes of tumors, Complete Genomics has tailored its sequencing service to do much of the computational analysis to highlight the differences compared with the standard reference genomes that are publicly available, Reid says. Entire long stretches of DNA can get wiped out, or switched around in reverse order, he says.
Here’s how the order flow is basically supposed to work, as I gathered from Reid. A researcher may take a biopsy of, say, a prostate tumor along with an ordinary blood sample from the patient. That researcher sends the two samples in for complete genome sequencing, the company assembles the raw data, and highlights the variations it sees compared to public reference genomes, and identifies which variants are known to be associated with disease. The data gets uploaded to remote cloud computing servers operated by Amazon Web Services, which enables the researcher to download the data via the Internet. Often, the researcher asks Amazon to copy the data on a disk and send it to the lab through commercial shippers like FedEx. Then the researcher is supposed to be able to use his or her creative brain to look at the data and gain the kind of insights that get published in Science, Nature, or the New England Journal of Medicine.
The process—from the moment the sample gets shipped to Complete Genomics to the moment data gets sent back—is supposed to take 90 to 120 days, Reid says.
“We provide the data so that it’s research ready,” Reid says. “You shouldn’t have to take on an enormous amount of data analysis to get to the biologically important stuff.”
It’s still way too early in the game to really say how well Complete Genomics is doing with this approach. It started offering the commercial service in a serious way in the third quarter of this year, when it shipped 300 complete genomes to customers, Reid says. The company has now established the necessary technology and infrastructure in Mountain View so that it can sequence as many as 400 complete genomes per month, or roughly 5,000 through all of 2011.
So far, Complete Genomics has signed up 35 customers, including the National Cancer Institute, the Institute for Systems Biology, the Broad Institute of MIT and Harvard, Genentech, Eli Lilly, and Pfizer, according to a recent regulatory filing. Most of the orders start out small, Reid says, asking for a handful of genomes. Some customers have made repeat orders, and the company’s goal as it builds up a sales force will be to win more repeat business, and more big orders, Reid says.
The company had built up a backlog of orders for 800 complete genomes as of Sept. 30, which represents an estimated $9 million in orders. Reid didn’t offer an update on how big the backlog is today.
The market is still full of guesstimates, because it’s hard to say how researchers will want to use all this genomic data. PacBio, in its IPO prospectus this fall, said the DNA sequencing market is expected to grow from $1.2 billion in 2009 to more than $3.6 billion by 2014, citing figures from Scientia Advisors, a market research and product development consulting firm.
Complete Genomics isn’t providing a financial forecast for 2011 for investors, Reid says, because the market is still at too much of an unpredictable stage. While his company bled lots of cash during its technology development years—$108 million from inception through June 30—the burn rate is going to come down now that the company is generating its first real revenues, Reid says. The $54 million in IPO proceeds will last at least through 2011, he says.
There are three big hurdles Complete Genomics needs to clear in its early days in the marketplace, Reid says. It needs to reduce the number of errors in its datasets, continue to bring down the cost, and boost its productivity so it can increase capacity.
Right now, the company’s genomes come with errors once every 100,000 data points, meaning that there are about 30,000 mistakes in each genome. “We are working very hard to reduce the errors to virtually zero,” Reid says.
The average cost per genome is still hanging around $10,000, although Complete Genomics does offer discounts for high-volume customers, Reid says. The market will open up even more when the cost comes down a bit further, he predicted. Reid noted that the BRCA1 and BRCA2 diagnostic tests for breast cancer cost around $3,000, and complete genomes will have to get down to that level and below to make genomes “a ubiquitous medical tool.”
The third thing is capacity. There is potential to increase the productivity of the company’s existing technologies, and buy more high-throughput machines, so long as there’s revenue coming in or investment capital to support that kind of expansion.
Both Complete Genomics and PacBio have seen their shares decline in their early days, which doesn’t exactly suggest that we’re living in a Genomics Bubble 2.0 in which they will have limitless budgets with which to dream big. But almost on a daily basis we are seeing startups—like San Francisco-based Sequenta yesterday—that would have been truly impossible to start in the era of high-cost DNA sequencing, of, say, two years ago. I asked Reid what kind of interest he saw on this fall’s IPO road show, and at least to hear him talk, there’s more enthusiasm than can be seen on his new public stock ticker.
“There is enormous interest in human genome sequencing as people realize. The people in the investment community that we talk to are fully cognizant of what a game changer this is,” Reid says. “It’s a key driver of personalized medicine and the entire diagnostics industry, and one of the key changes that will happen throughout healthcare for coming decades. The interest in this category of technology couldn’t be higher.”