Since 1986 - Covering the Fastest Computers in the World and the People Who Run Them

Language Flags
October 15, 2008

Complete Genomics Takes Off

Tiffany Trader (HPC)

Last week, San-Francisco-based Complete Genomics came out of stealth mode to become the first provider of large-scale human genome sequencing services. They claim to offer a third-generation genome sequencing technology that generates genomic data at a higher throughput than existing approaches and at lower cost.

What makes Complete Genomics different is that they are offering human genome sequencing as a service through their commercial-scale genome center. The technology and business model combine to enable large-scale human genomic population studies, thereby providing the basis for significant genomic analysis. By gathering and analyzing a large amount of genomic data, an individual’s genetic profile data can be applied to disease prevention and management.

HPCwire recently asked company representatives to share some details about their work. Complete Genomics Chairman, President and CEO Dr. Clifford Reid and Vice President of Software Bruce Martin took the time to respond.

HPCwire: Can you describe the sequencing service and talk about the practical significance of its use in health care?

Dr. Clifford Reid: Complete Genomics is offering the industry’s first large-scale human genome sequencing service for $5,000 per genome. We plan to sequence 1,000 complete human genomes in 2009 and 20,000 genomes in 2010. For the first time, companies and research institutions will be able to run large-scale complete human genome studies to understand the genetic basis of disease and drug response.

HPCwire: Can you provide a brief description of the technology pieces that make this sequencing service possible?

Reid: Complete Genomics has developed two breakthrough technologies that enable us to offer complete human genomes for $5,000. The first is a new method for creating extremely high density DNA arrays, which dramatically reduces the reagent and imaging cost of DNA sequencing. The second is a new ligation method of reading DNA, which dramatically reduces the reagent cost while maintaining the high accuracy of ligase-base DNA sequencing.

HPCwire: What is unique about the business model that allows you to do this?

Reid: By selling services rather than instruments, Complete Genomics is able to eliminate the burden of purchasing and operating complex and expensive DNA sequencing instruments, and eliminate the burden of building and operating a high-performance datacenter.

HPCwire: How big do you think the market is for your service?

Reid: Complete Genomics believes the market for large-scale complete human genome studies will be $3-5 billion in five years.

HPCwire: What would prevent competitors from copying your approach?

Reid: Complete Genomics owns or has licensed 110 patents and patent applications worldwide to protect our technology.

HPCwire: You quote a $5,000 price tag on the service for one genome in Q2 2009. What is the cost of a complete human genome sequence today?

Reid: Complete Genomics sequenced a complete human genome in July for $4,000 materials cost — that does not include equipment, labor, or overhead costs. When we launch our commercial service in Q2 2009 we expect our materials cost to be under $1,000 per genome, and our $5,000 price will cover all of our costs.

HPCwire: Focusing on the computational aspect of the service: What specific types of compute and storage system or systems are being employed to analyze the genetic data?

Bruce Martin: Complete Genomics uses a high-performance computing cluster, built using commodity servers (currently with Intel CPUs), Linux and other open source platform software, and clustered NAS systems.

HPCwire: What software is being used to do this and what is its source – proprietary in-house, commercial ISV, or open source?

Martin: Our system has an an open source platform with Linux, Sun Grid Engine and other management/operations tools. And we have proprietary applications for data analysis, base calling, alignment and assembly, which are all built in-house.

HPCwire: What percentage of the total expense of the infrastructure is represented by the computational infrastructure?

Martin: Computing is roughly 50 percent of the cost today. We expect that to decrease as a fraction of total cost over time.

HPCwire: Do you have HPC expertise in-house to help with the management of the compute resources and the data analysis? If so, could you briefly describe that expertise?

Martin: Yes, Complete Genomics has an experienced multi-disciplinary team, with world-class expertise in bioinformatics, in particular image processing, base calling, alignment, assembly and other areas of sequence analysis. We also have extensive knowledge of high-scale scientific computing, including monte carlo simulation, machine learning, and graph-based algorithms. Other proficiencies include large data set search and indexing, and datacenter operations and management.

HPCwire: Are you counting on projected increases in computational power to drive the growth strategy (1,000 genomes in 2009; 20,000 in 2010), and/or are you also intending to increase computational infrastructure? Or are there other pieces of technology or the business model that you intend to ramp up over the next couple of years?

Martin: Both — Complete Genomics will deploy significantly more infrastructure, and we plan to do so in a modular manner, thereby taking advantage of improvements in basic computing technology, for example, larger/faster disk drives, and faster and lower-power CPUs.