Sequencing

The target of $1,000 for a complete individual DNA sequence—an individual’s genome—has long been touted as a Holy Grail, an amount assumed to be affordable to individuals at least in the rich countries. Whether that’s true or not, absent market manipulation there is no reason for sequencing cost to stabilize at $1,000. Quite the opposite is likely, to the point where every individual on the planet could be sequenced. The reason is that the technology grows exponentially more powerful.

The Exponent
In August, a Stanford University professor claimed to have sequenced his entire genome for less than US$50,000 with a team of just two other people, in a single lab, with just one sequencing machine. A great effort, but the professor was behind the exponential curve. Two months earlier, genomics technology company Illumina launched a $48,000 genome-sequencing service, and two months later (October), Illumina was overtaken by Complete Genomics, offering a full sequence for $20,000—currently the cheapest available anywhere.

Complete Genomics also said it would have the price down to $5,000 in 2010 or 2011, a degree of confidence seemingly justified by more than just the exponent: The company recently said it has sequenced three human genomes at an average fixed cost in chemicals of $4,400; the last of the three costing only $1,726. Adding to this the cost of the sequencing machines, computers, human labor, and profit margin brings the price up to the current $20,000 tag, but these variable costs would be much lower given volume.

Here’s how the exponent looks today:

One reason why we are confident that $5,000 or even $1,000 is not the end of the exponential story is that several groups are developing “nanopore” chips, which will eliminate the time- and money-consuming need for chopping, amplifying, labeling, and re-assembling short stretches ofDNA, which is how sequencing is done currently. The intermediate result of the current method is a digital jigsaw puzzle of 3 gigabits of chopped-up data needing re-assembly before it can be analyzed. In contrast, nanopore chips can handle much longer stretches ofDNA.

IBMis using one of its Blue Gene supercomputers to simulate nanopore chips and run virtual experiments with them in silico. The computer can currently calculate the physics of tens of thousands of atoms per picosecond (trillionth of a second) in the DNA molecule and is expected eventually to model 200,000 atoms per picosecond. When translated from the virtual to the real world and fabricated in large arrays, the chips will have the ability quickly and cheaply to sequence genomes of individuals and greatly accelerate the trend to personalized medicine (medicine tailored to the patient’s individual genome.)

Oxford Nanopore is tackling the same problem of reading long sequences, having already demonstrated that its technology can identify DNAbases “with near total accuracy,” reading theDNA directly, without the use of costly chemical labels and imaging. The system can also detect, much more easily than current sequencing methods, epigenetic (environmentally caused) changes inDNA, which can result in disease including cancer.

IBM’s large arrays could perhaps make Oxford’s technology scalable and commercially viable, but meanwhile Oxford has licensed technology for nanopore sequencing methods developed by others, in hopes of becoming the first to meet or beat the US National Human Genome Research Institute’s target year of 2014 for successful nanopore sequencing.

Genetic Tests

It should be noted that Illumina only sequences theDNA, leaving the more difficult clinical interpretation to companies such as Knome, 23andMe, Navigenics, and Decode, who already use gene chips (microarrays) to analyze specific genetic variations for disease predisposition. Whole-genome sequencing captures a much higher volume of genetic information, as well as additional types of genetic variation such as deletions and duplications of segments of the genome.

Illumina apparently intends to compete indirectly with these companies, however, by developing an iPhone app enabling consumers to analyze their genetic information and receive guidance on (for example) correct medications and doses. The cell phone is becoming the medical diagnostic device-equivalent of the Swiss Army knife.

How Good Are They?

How valid and reliable are the testing services such companies offer? In the cases of 23andme and Navigenics, superstar geneticist Craig Venter and colleagues set out to find out. Each company bases its interpretation of the raw sequencing data on published findings of genetic markers for disease. Venter et al. found that for the five patients and seven diseases tested for the comparison, the raw data supplied by each company were almost 100 percent consistent.

But their interpretation of the data was less so. For instance, for lupus and type 2 diabetes, three of the five patients received conflicting results from the two companies. For psoriasis, 23andme reported a risk factor of 4.02 (four times greater) for one individual, while Navigenics reported only a 1.25 risk factor (25 percent greater), a threefold difference.

Technology Review contributor David Ewing Duncan compared results from deCodeme as well as from 23andme and Navigenics, and found that for overall risk of heart-attack, Navigenics reported high, 23andme medium, and deCodeme low. He also found several (but less-striking) contradictions for diabetes, macular degeneration, and other diseases.

23andme and Navigenics have agreed with the Venter’s recommendations on ways to reduce their inconsistencies, and while they would like to jointly “co-develop [voluntary] standards for the field,” they seem to recognize the need for a neutral party such as the Centers for Disease Control, the Food and Drug Administration, or the National Institute of Standards and Technology to set the standards.

Venter does acknowledge that advances in sequencing technology is rapidly producing a more accurate and thorough alternative to the those currently used by these companies.

Gene Chips Better than Karyotyping for Unborn Screening

If nanopore sequencing makes current methods of detecting genetic abnormalities—fast and accelerating though they are—begin to seem old and clunky, it makes the current method of screening unborn children for genetic abnormalities such as Down syndrome seem positively Stone Age. That method, called karyotyping, tries to identify problems by examining the size and shape of chromosomes. Such a gross test misses many serious conditions which thus go undetected until birth.

Baylor College of Medicine researchers have found that a type of DNA chip used in pediatric medicine, called “array comparative genomic hybridization” (aCGH) chips, can identify more than 270 genetic syndromes, provide a more detailed and accurate genetic profile, and reliably detect far smaller chromosomal abnormalities of a fetus than karyotyping.

Out of 300 cases examined in the Baylor study, aCGH testing provided new information about the risk of disease for seven of the cases, including two cases that would otherwise have been missed. However, aCGH testing is more expensive than karyotyping, and both require invasive procedures to extract amniotic fluid or placental tissue for testing, which can result in miscarriage. Several research teams are currently working on ways to obtain stray fetal cells from the mother’s blood, instead.

Cardiovascular Genetic Test from deCode

DecodeMe (a division of Decode) launched two new tests in January this year: One detects eight genetic variations associated with the risk of heart attack, intracranial and abdominal aortic aneurysm, stroke and atrial fibrillation, peripheral arterial disease, and venous thromboembolism (clots in blood vessels). The other test—deCODEme Cancer—measures 29 variations associated with the risk of prostate, lung, bladder, colorectal, and breast cancers, as well as basal cell carcinoma.

At $195 and $225, respectively, the tests are cheaper than Decode’s $985 genome-wide screen, which assesses genetic risk for 34 diseases and traits ranging from diabetes to male-pattern baldness. The variations detected by the Decode tests have been shown in the research literature to increase only modestly an individual’s risk of the diseases with which the variations are associated, so they are not all that helpful in determining preventive or therapeutic interventions. However, it is important for a physician to know if a patient goes from no risk to moderate risk of a disease, and the tests may at least reveal that much.

As understanding of the combinatorial effects of multiple genetic variations grows, the predictive power of the tests will increase. And at the very least, the data gathered through such testing are very useful in research into the diseases themselves.

Blood Test for Colon Cancer

A Tel Aviv University researcher has developed a simple blood test to detect a biomarker for polyp precursors to colorectal cancer. A press release issued by American Friends of Tel Aviv University stated: “This painless, non-invasive and inexpensive test could very well be a breakthrough of the decade.” While “not 100% accurate,” detection of the polyp biomarker should be enough to persuade patients to undergo colonoscopy, which is expensive and can be painful and they might otherwise avoid. It should also enable doctors to prescribe preventive therapies and lifestyle changes.

The test is being commercialized by Bio Mark Ltd., a subsidiary of Micromedic Technologies Ltd.

Detector for Multiple Biomarkers

A prototype compact detector uses nanoscale magnetic beads to tag and identify multiple cancer biomarkers simultaneously in a human blood serum sample, with sensitivity tens to hundreds of times higher than current single-marker detectors, which use fluorescent tags.

The detector contains a silicon chip with 64 embedded sensors whose electrical resistance changes in the presence of a nearby magnetic field. In a test involving seven potential cancer biomarkers, concentrations ranging from 5 quadrillionths to 0.1 trillionths of a mole (a standard unit of measurement for molecules) were unambiguously detected simultaneously.

A company has been formed to develop and commercialize the detector.

Toward a Handheld DNA Diagnostic Device for Doctors’ Offices

The four-year EU-funded SMART-BIOMEMS project completed earlier this year developed a prototype microfluidic chip intended to be the heart of an inexpensive portable diagnostic device to simultaneously and automatically analyze multiple DNA samples. The prototype was designed for cancer testing and diagnosis. It is by no means the only such project, and the project’s website is silent about the final result.

What To Do With All the Data?

Genome + EMR

Brigham and Women’s Hospital (BWH) plans to collect blood samples from all consenting patients, so that patients’ genomes can be analyzed together with clinical and demographic data contained in the hospital’s electronic medical record (EMR) system. The resulting massive EMR+genome database will allow physicians and scientists to research genetic risk factors and predict how well a therapy will work for the individual patient.

A pilot project involving 600 patients is about to be launched, to determine patients’ willingness to enroll, how best to explain the project to potential participants, and how to minimize disruption to physicians’ work flow and patients’ visits.

BWH is not alone. Over the next two years, Kaiser Permanente researchers plan to conduct genetic analyses of 100,000 older Californians and put that genomic data in the context of detailed patient data from electronic health records, patient surveys, and records of environmental conditions where the patients live and work, in order to uncover the relationship among genes, the environment, and disease. All of the resulting information will be made available to other researchers through the US National Institutes of Health. The overall goal is to advance personalized medicine and more efficiently utilize healthcare resources.

The patients’ DNA, obtained from saliva samples, will be analyzed for 700,000 genetic variations using Affymetrix gene chips.

A similar project in the UK, the UK Biobank, is in the process of collecting samples from 500,000 people, but genetic analyses have not yet begun. The Mayo Clinic has also begun to create a biobank of genetic information on 20,000 patients plus several smaller banks focusing on specific diseases, including bipolar disorder. Eventually, the various biobank projects may be linked into one large database.

Epigenetics

Having available the genome plus the clinical, demographic, and family history data increasingly stored in the EMR and PHR (personal health record) will be of tremendous help to researchers in another vital area of genetics/genomics: Epigenetics.

Lamarck proposed a version of evolution whereby acquired characteristics are passed on to offspring. Largely forgotten in the shadow of Darwin’s competing theory of natural selection, Lamarckism has received some partial corroboration from a study showing that the offspring of mice genetically engineered to have memory problems, then raised in an enriched environment (toys, exercise, social interaction) for two weeks during adolescence, had better memory despite having the genetic defect and no exposure to the enriched environment.

In a second study, rats raised by stressed mothers that neglected and physically abused their offspring showed specific epigenetic modifications, grew up to be poor mothers, and appeared to pass these changes on to their daughters.

Such “epigenetic” changes (inheritable changes that change the expression but not the sequence of genes) play a major role not only in development but also in disease.

Rapidly advancing technologies such as DNA microarrays and DNA sequencing facilitate the study of epigenetic changes linked to environment and behavior. Still, the task is daunting: “A human epigenome project would be the equivalent of 250 human genomes,” one researcher told Technology Review’s Emily Singer. But it would be worth it, because these recent findings have very profound implications. For instance, they suggest that girls’ education, too often neglected or even discouraged in some countries, is important not just to the present generation of girls but to their eventual children.

Epigenetic Memory Boost Therapy

Existing drugs that inhibit enzymes that regulate gene expression could enhance learning in both normal mice and those that are cognitively impaired. In 2007, brain-damaged mice given such an inhibitor drug were able to recall lost memories. EnVivo Pharmaceuticals is developing more potent inhibitors that can easily enter the brain, and as of last December was hoping to start clinical trials this year.

Epigenetic Treatment for Acute Lymphoblastic Leukemia

A reversible epigenetic change has been discovered that could lead to a cure for children with a subtype of acute lymphoblastic leukemia (ALL) known as MLL-AF4, which accounts for 5 percent of all ALLs but 70 percent of ALLs striking infants. Children with MLL-AF4 have only a 50 percent chance of cure with chemotherapy.

The discovery, first made in a mouse model of MLL-AF4 and then validated in human patients, was that the abnormal “fusion” protein that characterizes the disease changes the cell’s DNA. It appears that an enzyme necessary to the change process could be fairly easy to target with small-molecule drugs. Some drugs that target similar enzymes are already in use to treat a kind of lymphoma and are being tested in other cancers.

Next Month

In the December Digest, we will report on progress in gene therapies.