How To Decode A Genome In Just A Single Day

The Harvard/ MIT Broad Institute and Intel set a new speed record for analyzing genomes.

Genome analysis is big business. Thanks to the human genome project, two related fields of medicine—computational biology and personalized medicine—have become part of our health care vocabulary over the past few years. Because genome sequencing requires multi-terabyte files and mind-boggling amounts of server power, it's been a boon for tech companies. One of them, Intel, announced this week that they can now analyze a genome in only one day, a 300% increase in speed from the previous three days it took for genome decoding.

Intel's new benchmark was announced in conjunction with the joint MIT-Harvard Broad Institute. In recent years, Intel has aggressively targeted customers in the health care industry (along with automotive-sector customers) in order to build alternative profit bases. Industry trends like the rise of tablets and smartphones, along with the rising enterprise use of cloud computing have hurt Intel's traditional business model; health care gives Intel a lucrative new industry to work in.

Eric Dishman, an Intel health care executive and one of Fast Company's Most Creative People, said that the work with Broad was part of a much larger effort in North America and Europe to develop a health care tech infrastructure that would "take genomics to the bedside and to the clinic" rather than only research hospitals. Dishman has an unusual resume for a tech executive; he was diagnosed with kidney cancer in college and credits genomics with helping to save his life. And 23 years after receiving his diagnosis, a genomic analysis led doctors to begin treating him with a drug intended for another sort of cancer. This drug treatment allowed him to become healthy enough to receive a kidney transplant.

The increase in genome processing speed comes through a new version of Broad's Genome Analysis Toolkit (GATK) running on Intel servers. According to Broad and Intel, the new benchmarks in variant discovery don't just cut analysis times from three days to one day, but can also analyze data sets with 100 times more DNA samples than previously possible. These changes are due to both the new software update and the use of instruction sets related to highly compute intensive tasks that were released in late 2013.

Other Intel partners in the genomics sphere include London's Francis Crick Institute and the Beijing Genomics Institute. Dishman added that "(We work) on the technical challenges. We need to make the computer as fast as possible and then there's the storage. This is some of the biggest of big data, if you generate a file every time a patient goes to the clinic, someday that will be a huge problem." The end goal? To make the Broad's genomics programs "hum like a well-tuned engine."

Add New Comment

0 Comments