Intel and the Oregon Health & Science University are teaming up on a supercomputing project to speed up analysis of human genetic profiles, which could help with personalized treatment for cancer.

The goal for the multiyear partnership is to make genetic analysis a routine part of patient care.

“We are generating an enormous amount of data, which has historically been unmanageable,” said Joe Gray, a research director at OHSU’s Knight Cancer Institute, in an interview.

Gray envisions that the project will involve a “lot of back and forth” between geneticists, engineers, and biomedical scientists before they can map the human genome to identify the mutations that lead to cancer.

Such a map would provide a better understanding of an individual’s genetic makeup, so biomedical engineers can cancer treatment that kills only the mutating cells, not the healthy ones.

It sounds simple, but Gray’s project is grappling with delivering the right kind of computer power for a project of this scale. “We are talking about a terabyte of data per patient that describes the molecular architecture of a disease,” he explained. If this is extended to millions of patients a year, it’s an awful lot of data.

Researchers will use Intel’s Xeon E5 HPC CPU, which offers Intel Trusted Execution Technology (TXT) and Intel Node Manager Server power-management technology.

Aside from the technology, which is steadily evolving to manage this volume of data, the other challenge is an ethical one.

“Privacy is a clear issue, and the data needs to be handled confidentially,” said Gray. When asked about whether the data will be stripped of any personally identifiable information, Gray admits to another “gray area.” Our DNA is like a molecular blueprint — theoretically, it is possible to match and reestablish a patient’s identity without having a direct link.

But the partnership with Intel is designed to explore the architectures best suited to deal with this data, and tackle ethical issues at a later date. The more immediate goal is to identify a number of challenging biomedical projects, and to experiment with hundreds of hardware and software packages.

An example of a project is tracking a typical cancer tumor through its thousands of genomic changes. “Some of them are important in the way that the disease responds to therapy,” said Gray. The team intends to build computer algorithms to predict how best to manage the disease as it progresses.

This isn’t the only partnership between a technology company and medical institution; Illumina and Life Technologies Corp recently announced rival Boston-based clinical genomics initiatives.

“Our joint team plans to do research not only into the mutations but also the ‘circuitry’ that enables malignant cells to spread,” Intel’s general manager for health care Eric Dishman, noted in a blog post. “The ultimate hope here is to learn how, for a specific individual, this circuitry can be “turned off” to stop the spread of cancer cells.”

Dishman, recently had his genome sequenced when recovering from a kidney transplant.

“It, too, took weeks of computing and then months upon months of analysis to make sense of my own unique case,” he concluded. “Today, these tools are too slow, too expensive and too rare—I want to make sure everyone has access to the kind of customized care that I lucked into.”

Curious to learn more? We’ll be inviting world leading geneticists to delve into the tricky, ethical issues at HealthBeat, our health care conference on May 20 & 21 in San Francisco. 

Scientific researcher image via Shutterstock

As the cost of DNA sequencing continues to fall, entrepreneurs and investors see an opportunity to bring bioinformatics tools to a mass market.

“Innovations in DNA sequencing have led to an explosion of data,” said Rachel Pike of venture firm Draper Fisher Jurvetson (DFJ), adding that the developments will have “real and lasting implications” for drug development, as well as the biological production of chemicals and fuels.

Until recently, it has been costly and time-consuming to map the 3 billion units of DNA, known as base-pairs, that make up the human genetic code.

To bring gene sequencing to a mass market, DFJ is stepping up its investment in the space. Today, the storied venture firm funded a company that processes genetic information for the purposes of clinical medicine and research. The Seattle-based startup, “Spiral Genetics,” has raised $3 million to build out tools that help researchers process and analyze genomics data.

“Our CTO was looking at a lot of the bioinformatics programs available that had been created by the open source community,” said CEO Adina Mangubat of the company’s early days in 2009. “We all realized that in a few years there would be a huge need for fast scalable bioinformatics tools.”

The company charges based on the amount of data analyzed. Customers can either purchase ‘pay as you go’ credit packs or sign up for an annual subscription. Mangubat would not disclose any customers but said the cloud-based technology is currently used in myriad use-cases from selective corn breeding to childhood cancer diagnostics.

Mangubat is confident that we are already experiencing the benefits of the explosion of data and gene sequencing technologies. Already, the space is getting flooded with new technologies — Spiral Genetics competes with Knome and DNAnexus.

To push into the medical research field, Spiral Genetics also announced that it has secured a partnership with Omicia, a provider of technology that is used by researchers and clinicians to analyze genomes and prioritize disease-causing variants.

Genomics entrepreneurs and researchers believe the low cost of human DNA sequencing is the most exciting development since the completion of the Human Genome Project.

“It’s no longer business as usual in medicine,” said Dr. Dietrich Stephan, cofounder of Navigenics, in a recent interview. Patients will be “touched and informed by genetics” in the next 2-5 years, he said, and hospitals will increasingly adopt these gene-sequencing technologies to better treat and diagnose disease.

This guest post is written by independent business transformation consultant Theo Priestley.

Remember Johnny Mnemonic? It was the William Gibson short story-turned sci-fi film where Keanu Reeves carried large packages of data in his head. Far-fetched?

Not anymore. Boffins in the UK published results this week showing that DNA can be used to archive vast amounts of data. One gram of DNA can hold the equivalent of two petabytes of data. In the report published in Nature.com they encoded a clip of of Martin Luther King’s classic address from 1963, a JPEG photo, a PDF of the famous Crick and Watson paper describing the structure of DNA, a text file containing all of Shakespeare’s sonnets and a file about the new encoding system itself, in total 738kb of data onto DNA, then sequenced and reconstructed the original files with complete accuracy.

According to the report, the theoretical limits of data storage on DNA go way beyond what is physically possible using today’s methods. A DNA-based storage system requires no maintenance, no electricity and no backward compatibility requirements to retrieve the data. As long as there is DNA based life on Earth there will always be a means to read it, says Dr Ewan Birney of the European Bioinformatics Institute who co-authored the report. To copy data to DNA the team translates the binary into a bespoke code which a synthesis machine then resequences as DNA.

It’s not the first time this has been achieved, but the techniques differ from those used by a US team in Boston who encoded an entire book using their methods, the results of which are also published in Science Magazine. DNA is held together by four chemical groups and the UK team’s storage system uses those same four groups or “letters” but encoded in a completely different language to the one understood by the building blocks of life. But where it differs is in how it’s treated as not one long molecule, here the UK team created multiple copies of overlapping DNA fragments, with each fragment also carrying some “indexing” information that identify where in the overall sequence it should sit. According to the report this builds in redundancy into the storage system, meaning that if some DNA fragments become corrupted the information is not lost.

This opens up massive possibilities for enterprise and government data archival using a molecular-based storage method say the scientists.

The downsides?

At the moment, it’s phenomenally expensive. But the report argues that as technology matures, the cost for entry will be lowered. But perhaps the biggest kicker is because the DNA is synthesized it can’t be placed inside a living host as it’ll be rejected and disposed of naturally by the body. So dreams of making a career from carrying storage for large corporations in your fingertip will have to wait for now.

Theo Priestley is a consultant, analyst, and advisor. He’s written analysis on the industry and tech space in general since 2007 and has collaborated with and advised the large and the small, from stealth startups to industry established players.

Cigarettes, chemicals, and sunlight are well-known causes of cancer, and people can respond accordingly to these risk factors by abstaining from smoking, buying organic, and slathering on the sunscreen. But what about the causes that can’t be mediated?

Genetics (and family history) play a significant role in cancer detection and treatment as well, and Foundation Medicine has raised $42.5 million for its technology that investigates the genetic makeup of cancer patients and uses this knowledge to match patients with the best treatments.

The flagship product, FoundationOne, is a genomic profiler that uncovers genetic alterations in a patient and reports them to the physician. With a deeper understanding of each particular case and the ways the cells are working, physicians can better prescribe drugs, clinical trials, and other medical regimens.

FoundationOne is available for all solid tumors, and results can be gleaned from just 50ng of DNA. The platform can easily be integrated into an oncologist’s practice and with this funding, will continue to develop and scale the product commercially.

Investors include Deerfield Management Company, Casin Capital, Redmile Group, Roche Venture Funds, and WuXi Corporate Venture Fund. Previous backers Google Ventures, Kleiner Perkins Caufield & Byers, and Third Rock Ventures contributed again, following last year’s Series A of $33.5 million.

Foundation Medicine is based in Cambridge, Massachusetts. Its mission is to create products and services based on genomic analysys that can be used in cancer diagnosis and treatment. Read the press release.

DNA is a fantastic medium from a storage perspective. Its storage density is one million gigabits per cubic millimeter. A gigabit is a billion bits, and one gigabit equals 125 megabytes. So a million gigabits, if I’m not such a math idiot as some have said, is 125,000 gigabytes … in one cubic millimeter.

That’s a little more storage than fits in your new MacBook Air.

The Harvard scientists followed a four-step process: encoding the book into the zeroes and ones of computer storage, and then into the language of DNA. To read the data, they then synthesized the DNA and decoded the gene sequences:

Source: Harvard

Storing data in DNA

The key, according to author George Church, is that “most non-DNA methods store on a plane what DNA stores in a volume.” That 3-D storage method is about as good as it gets: According to Sri Kosuri, one of the scientists who made it happen, the storage density compares “favorably with other experimental storage methods from biology and physics.”

In fact, it’s so good, according to a Harvard Medical School write-up, that all the data the world creates in a year could theoretically fit in four grams of DNA.

In addition, DNA is stable at room temperature, unlike other high-density storage methodologies such as quantum holography, which requires near absolute zero temperatures and massive energy expenditures. One downer: encoding is slow, and decoding is also.

And the book?

It may not exactly be bedtime reading unless you’re actually trying to sleep … “Regenesis: How Synthetic Biology Will Reinvent Nature and Ourselves” … but there are now more copies in existence, sort of, to make it far and away the biggest bestseller ever.

If only each one was paid for.

photo credit: kevin dooley via photo pin cc

You may not think music is in your genes, but a new technique from personal genomics company 23andMe may just prove you wrong.

Using the genetic information available in just a few drops of saliva, 23andMe has developed a way to create music out of your genetic code.

The general technique of creating music out of DNA isn’t anything new, but 23andMe composer Mark Ackerley says that his technique is a bit more advanced.

While previous techniques use DNA as a jumping point for a composer-created melody, the 23andMe method is almost fully automated. Pitch, key, rhythm — all of the most significant parts of a musical composition are assigned by genetic data.

To determine key, for example, 23andMe takes a user’s maternal haplogroup, a group of DNA sequences that’s used to determine ancestry, and links it to a specified key. Likewise, traits like height and eye color also correspond to their own musical values.

And while all of that may sound the recipe for a cacophonous, painful bit of music, the technique’s developers say that just isn’t the case.

“By using these guidelines we ensure we won’t get random notes or disjointed melodies,” Ackerley said on the 23andMe blog.

In all, Ackerly says that there are hundreds of thousands of possible melodies available via the new technique. You can find just one example below.

Researchers have previously been able to write data in DNA, but only once. Once written, the information was read-only. Now researchers are able to write and rewrite data. They’ve only done one bit at a time, but that’s an essential component of computation, and an important step forward in the potential creation of a biological computer.

“Now we can bring logic and computation inside a cell itself,” Jerome Bonnet, one of the scientists behind the innovation, said in a statement to Science News.

In early tests, the scientists essentially created a traffic light out of an Escherichia coli bacterium. The goal was to flip a sequence of its DNA back and forth: a way of storing information comparable to storing a zero or a one in a conventional computing bit of data. They choose a segment of the bacterium’s DNA that impacts color, so they could visually determine the experiment’s success or failure. Then, by flipping a small sequence of its DNA, they caused the bacterium’s cells to fluoresce in green. Switching it back, they caused it to shine red: visual proof that they were both rewriting DNA and storing information.

DNA has long been known to be an almost unbelievably efficient information storage medium. With 3.2 million base pairs in a strand of 20-25,000 genes, your DNA stores 800 gigabytes of data in a literally microscopic amount of space. Compare that with 50 gigabytes for your average Blu-Ray disc, or perhaps a terabyte for the hard disk in your computer — neither of which are microscopic — and you can see the appeal of learning to read and write DNA.

The scientists are calling it RAD, Recombinant Addressable Data, and have successfully saved data even during cell reproduction. The abstract of the paper detailing the innovation states: ”Our core RAD memory element is capable of passive information storage … for over 100 cell divisions.”

Even more importantly, the data can be re-written multiple times with no performance or integrity degradation. That’s critical for their long-term objectives: “extending computing and control methods to the study and engineering of many biological systems.”

While it’s unlikely that you’ll see numbers representing gigabytes of RAD data on your next computer purchase alongside RAM and ROM, future uses include potential monitoring and treatment of cancer. That’s something to get even more excited about than a multi-gigabyte RAD array.

Top image credit: Micahb37 via Flickr

New investor Johnson & Johnson Development Corporation and existing investors New Enterprise Associates and Google Ventures participated in the round. 23andMe plans to use the funding to accelerate research and development.

Based in Mountain View, Calif., 23andMe was founded in 2006 and has raised more than $44 million to date.

The WaPo’s Rick Weiss, one of the few mainstream reporters to tackle the consequences of the Encode report (which was published in 29 separate papers), put it this way:

The new work also overturns the conventional notion that genes are discrete packets of information arranged like beads on a thread of DNA. Instead, many genes overlap one another and share stretches of molecular code. As with phone lines that carry many voices at once, that arrangement has prompted the evolution of complex switching, splicing and silencing mechanisms — mostly located between genes — to sort out the interwoven messages.

The new picture of the inner workings of DNA probably will require some rethinking in the search for genetic patterns that dispose people to diseases such as diabetes, cancer and heart disease, the scientists said, but ultimately the findings are likely to speed the development of ways to prevent and treat a variety of illnesses.

One implication is that many, and perhaps most, genetic diseases come from errors in the DNA between genes rather than within the genes, which have been the focus of molecular medicine.

Complicating the picture, it turns out that genes and the DNA sequences that regulate their activity are often far apart along the six-foot-long strands of DNA intricately packaged inside each cell. How they communicate is still largely a mystery.

The implications are preliminary but profound, since so much of today’s cutting-edge medical enterprise is based on the premise that understanding genes and their variation is the key to understanding disease. Increasingly, that appears to be only part of the story — possibly not even a particularly large part. All of which suggests that understanding the genome wasn’t the beginning of the end of the quest to understand the workings of life — just the end of the beginning, and maybe not even that.

Scanning the genome of a DNA pioneer – Much as he must have anticipated, the news last week that James Watson, the co-discoverer of DNA, had sequenced his own genome drew a flurry of largely unenlightening media attention (see here and here for just two examples). Yes, it’s interesting to know that Watson has some of the same reservations about knowing his genetic disease risks as many others — the scientist didn’t want to know the status of his apolipoprotein E gene, which can indicate the risk of Alzheimer’s disease — and yes, as the technology improves, more people are going to want their genome sequenced. As the Encode study referenced above suggests, however, sequencing the genome is probably just the start of actually understanding an individual’s genetic makeup. So two cheers for Jim Watson and the low-cost genome sequencers, but no one should be under the illusion that this event is the “milestone” many have made it out to be.

Genetic-association overload – Scientists employing the technique of “whole-genome association” recently announced that common genetic variations appear to underlie seven common diseases — bipolar disorder, coronary artery disease, Crohn’s disease, hypertension, rheumatoid arthritis, and Type 1 and Type 2 diabetes. (See the NYT piece here, although a subscription may be required). The finding was noteworthy because it suggested that several of the diseases may share common origins — and, of course, because knowing the effect of these genetic variations might provide clues to new treatments.

This study is the latest of several that have recently turned up much more solid evidence of links between DNA variants and disease (see my earlier coverage here, here and here), and it seems safe to say that we’re just at the beginning of an avalanche of such announcements. Which, of course, means, it’s right about time for boredom to set in — and fortunately Tom Goetz is on hand to deliver. Now it’s time to anticipate the backlash.

Synthetic Genomics hits it big – At least in terms of valuation. As Matt reported yesterday, the synthetic-biology startup founded by genomics pioneer Craig Venter raised an undisclosed sum of venture funding and is now valued at something close to $300 million — according, that is, to Venter himself. Synthetic Genomics aims to create artificial microbes that could assist in the production of new clean-burning fuels — for instance, by converting coal into natural gas.

In a separate but related effort, Venter’s own research institute has been trying to determine the minimum number of genes necessary for life by systematically knocking genes out of a simple microbe. Earlier this month, a patent application from Venter’s institute claimed ownership of the 381 genes that resulted from this effort. The idea here is that it should be possible to synthesize that short genome, insert it into a microbe from which the DNA has been removed, and “boot up” a largely synthetic organism. The synthetic genome would be designed so that additional genes could be easily inserted, theoretically making it an ideal platform for industrial use. The patent application, in fact, claims production of ethanol or hydrogen fuel as an initial use.

What’s perhaps most striking about all this are the parallels to Venter’s early attempts to lay claim to large chunks of the human genome. (Those never really worked out, but not for his lack of trying.) As science writer Carl Zimmer points out in this post, Venter’s approach to synthetic biology seems to embody the same sort of land-grab mentality, by attempting to lock up the basic genes necessary for creating synthetic organisms. That stands in sharp contrast to the “open-source biology” movement, in which researchers are building publicly available “genetic toolkits” for designing and building new synthetic organisms.

In any case, it’s far from clear that Venter’s attempted land-grab will work any better this time around, but this could easily turn into another epic battle between “open” and “closed” technology philosophies. So make some popcorn and grab a seat.

Number of human genes finally determined? – One of the early conundrums created by the first human-genome map was the surprisingly small number of human genes turned up by the Human Genome Project. Although some initial estimates had ranged as high as 100,000 to 150,000, the first draft of the genome put the number at 30,000 to 40,000, and that number has been falling steadily ever since. (See here for details.)

Now, an MIT computational biologist named Michele Clamp has a new bottom-line answer: 20,488 genes. As Science‘s Elizabeth Pennisi writes in this news story (subscription required):

Clamp compared all the human genes in a database called Ensembl with those cataloged for dog and mouse. In all, 19,209 were the real, protein-coding McCoy, 3009 had been erroneously put on the gene list, and 1177 remained ambiguous, she reported.

She rated the “geneness” of these leftovers by comparing them to random stretches of DNA. Almost all made the grade with respect to a genelike proportion of the bases G and C, but not for features such as the distribution of short insertions and deletions in their sequences. Overall, 1167 were “bogus” and lacked any independent evidence that they coded for proteins, she reported. She did a similar analysis with the other gene databases, then summed the unique genes of all of them to get her final count.

Given the tremendous genome complexity that’s now coming into view, the low number of human genes isn’t quite the shock it once was — but it’s still nice to have an answer.