Virtual Children, Genome Sequencing for Everyone, and Forget Genetic Privacy
Ronald Bailey sends his first dispatch from the Fourth Annual Consumer Genetics Conference
Boston—The Fourth Annual Consumer Genetics Conference opened Wednesday drawing more than 300 participants from a wide assortment of companies and research institutes devoted to making personal genomics a reality sooner rather than later. On the first day, Harvard University genetics guru George Church argued that everyone should have all three billion base pairs that make up their genomes sequenced now, not later. Princeton University biologist Lee Silver believes that creating virtual genomes of future children can help parents make better reproductive decisions. New York Genome Center's Chris Dwan thinks that there will be no such thing as genetic privacy in the near future.
Generating Genomes for Virtual Children
Lee Silver, was the first keynote speaker, spoke about his just launched company GenePeeks, which computationally combines the genomes of potential mates to generate the genomes of virtual children. Silver illustrated the power of modern genomic science with the case of a friend whose brother and father had died of colorectal cancer. She had been tested for the most common gene variant for hereditary colon cancer, a dominant mutation (one copy will affect the carrier) in the MSH2 gene. Her physician at Cornell-Weill Medical Center had duly ordered the diagnostic test for that gene variant and reported back to her that it had only found a "variant of unknown clinical significance." She pushed the doctor to give her the genetic data which she then asked Silver to take a look at. It was clear from the family history that there might be a hereditary risk. A known mutation in the MSH2 gene knocks out a specific protein which leads to colorectal cancer. Silver put her data into the UCSC gene browser and did not find the known mutation, but instead found another which appeared to also knock out the crucial protein.
Silver's friend took this information back to her physician whose first response was that Silver was "playing doctor without a license" and had broken the law by making an illegal genetic diagnosis. Her physician added that the variant uncovered by Silver had never before been seen in the clinic and that the only way to find out that it is clinically significant is if someone becomes ill. Basically, the physician was telling Silver's friend that we'll know it's significant once you're diagnosed with colon cancer and not before.
Looking back Silver says that it was probably not a good idea for him to have done the analysis and that his friend probably should have talked with a genetic counselor. More disturbingly, his friend's physician refused to discuss the matter further with her, so she switched to a physician who would. "What he did was quite unethical," says Silver. "Her physician's response was deplorable." I completely agree. This episode simply underscores the point made by Columbia University pediatrician and geneticist Wendy Chung, that with respect to genetics, "Even if you have been out of medical school for five years, you are totally out of date."
GenePeeks will exploit the genetic knowledge that is steadily piling up in genetic data banks and scientific literature. Creating virtual children will give would-be parents some insights into the genetic glitches they may risk conferring on their real children. The company first plans to deploy its technology in conjunction with a sperm bank. The complete genomes of every donor will be sequenced and stored. When a woman seeks a donor, her genome will also be sequenced and then the two will be computationally combined to produce thousands of virtual children. GenePeeks will analyze these virtual genomes as a way to help a woman select a donor that offers the best chance of avoiding genetic problems with her children.
GenePeeks CEO Anne Morris, a Harvard University business professor, has a personal interest in the technology since her son Alec (now age 5) was born using donor sperm, and suffers from MCAD deficiency, a metabolic condition that prevents the body from converting fats to energy, particularly during periods of fasting. In order to prevent seizures, brain damage, or coma, MCAD children must be fed carbohydrates every three or so hours. The disease is autosomal recessive and about 1 in 70 people of European descent are carriers of the gene variant that causes MCAD. Compare this to the prevalence of the cystic fibrosis gene variant in which about 1 in 30 people of European descent are carriers.
Sperm banks currently advertise that they check donors to see if they are cystic fibrosis and sickle cell trait carriers and exclude them, but so far do not check for other disease gene variants. The GenePeek technology will solve this problem since all donors will be checked for known disease variants, but it offers even more information. Scores of genes acting together contribute to or reduce the risks for common maladies, like diabetes or heart disease. The genes that contribute to common disease risks will be computationally mingled such that clients will have some better idea of the relative risks of choosing some donors over others with respect to those diseases. The most likely outcome of going through this process is that clients will be steered clear of two to three donors among a hundred or so from which they could choose.
Direct to consumer genetic testing is already moving this direction. My 23andMe genotype screening test reports, of 48 gene variants associated with disease that the company tests, that I am the carrier of only one, alpha-1 antitrypsin deficiency (ATT). People with ATT are at much higher risk of emphysema and liver disease. Since my father had hypertrophic cardiomyopathy which is a dominant variant, the good news is that 23andMe reports that I don't have a gene variant associated with familial hypertrophic cardiomyopathy. But I already more or less knew that since I had had myself checked out after my father was diagnosed.
Part of the reason that GenePeeks creates virtual children is that it is a way to get around Food and Drug Administration medical device and diagnostics regulations. All such devices and tests must be FDA approved before they can be used to diagnose illnesses in patients. In this case, GenePeeks will not be interpreting the genomic data (arguably diagnosing) of either the donor or the recipient, just that of virtual children who do not exist.
Getting Genomic Information from Lab to Consumer
Next up was Michael Christman from the Coriell Institute for Medical Research, whose Personalized Medicine Collaborative has enrolled 6,000 participants. The Institute's approach feels very old-school, i.e., doctor knows best. A central committee of physicians, genetic counselors, geneticists, and bioethicists decide what information is "actionable" and thus is allowed to be doled out to participants. One interesting innovation pioneered by Coriell is that new findings are broadcast to all participants and they can decide for themselves whether or not to look at them. Apparently some people would simply rather not know more information about their risks of diseases like Alzheimer's or breast cancer. Christman made the point that in the very near-term the most useful genetic information will help guide physicians in selecting appropriate drug treatments for patients, e.g., 30 percent of heart disease patients do not respond to the anti-clotting medication Plavix so they could instead use Prasugrel instead. For what it's worth, 23andMe suggests that I would respond to Plavix should I need it.
Christman did persuasively outline an information ecosystem for handling genomic information in the future. The digitized results of whole genome tests done in gene sequencing "factories" would be sent for permanent custodial storage in GeneVaults. These data would be made available to third party interpreters—geneticists, disease specialists—who would deliver an actionable report to the physicians who initially ordered the sequencing tests. Since the data are permanently stored specific, gene sequences can be ordered up later to further interpretation.
The Genome and Big Data
The next panel discussion focused on how to handle and interpret the vast amount of data that whole genome sequencing for millions of patients will produce. Sandy Aronson of Partners Health Care Center for Personalized Genetic Medicine offered the intriguing idea that it isn't necessary to store all of a consumer's entire genome, just the 3 million or so gene variants that differ among individuals. Aronson and his colleagues have devised a nifty system in which doctors are automatically alerted when new research comes in regarding genes for which they have already tested in individual patients. For example, new research may determine that what was first identified as a variant of unknown significance is actually pathogenic and the physician can now address that issue in all her patients with that variant.
Chris Dwan, acting senior vice-president of IT at the new New York Genome Center, discussed how to handle the Big Data issues involving whole genome sequencing. Dwan was essentially optimistic that most of the infrastructure for storing and "slinging" vast quantities of genomic information already exists. For most purposes, Dwan argued that researchers and genomic data consumer services companies can store their information in the "cloud" with Amazon and the like and current high-speed internet connections are sufficient for getting data from one place to another.
Referencing Mount Sinai Medical School geneticist Eric Schadt's recent commentary in Nature, Dwan argued that as whole genome sequencing costs fall to near zero, trying to keep genomic information private will be futile. Unfortunately, our unsettled political climate could result in laws and regulations with regard to handling genomic information written in response to fear, uncertainty, and doubt. Such ill-advised privacy regulations would severely slow progress in turning genomic discoveries into effective treatments. Although I am far from persuaded by his argument, Dwan suggested that since genetic privacy will simply not be possible, laws against genetic discrimination similar to the Americans with Disabilities Act might be necessary.
The next panel dealt with a lot exciting developments in molecular diagnostics, but I will single out Sequenom's non-invasive test for Down's syndrome and other similar diseases. Typically everyone has 23 pairs of chromosomes for a total of 46. Down's syndrome occurs when there are 3 versions of chromosome 21. The new Sequenom test takes advantage of the fact that fetal cells shed into a woman's bloodstream where they can be removed and tested to see if the fetus she is carrying has Down's syndrome at about 10 weeks into her pregnancy. The new test is 99 percent accurate.
What's the Best Way to Sequence a Genome?
The next panel was the battle of the sequencers featuring presenters from Life Technologies, Complete Genomics Inc., Nabsys, Inc., and GnuBio. It's pretty clear that there's far more than one way to skin a genome. Life Technologies uses the Ion Torrent chip reader and will release its Proton chip next year which can decipher 1.2 billion DNA base pairs in hours. "The $1,000 genome is actually already here," declared Life Technologies vice-president John Stark. Rival sequencer Complete Genomics has built a sequencing factory that can decode 1,000 genomes per month. The company has already sequenced more than 7,000, according to chief business development officer Robert Klein. Klein just got his complete genome last week and has been delving into it. He admitted to being a bit disturbed when he uncovered a variant that looked like it might confer some risk of Huntington-like disease. Fortunately, further research suggested that his variant was not pathogenic.
Amusingly, Barrett Bready, CEO of Nabsys, Inc. began by citing a 2010 Reuters article, the headline of which asked, "Did the Failure of Genomics Doom the U.S. Economy?" Probably not, but perhaps new developments in genomics will pull us out. The Nabsys sequencing technology can read 1 million DNA base pairs per second. Finally, John Boyce, the CEO of GnuBio described how his company sequencing system based on emulsions using standardized cartridges can undertake 100,000 reactions per second. However, mentioned by many in the hallway discussions and hovering over the panel was the specter of the absent Oxford Nanopore which promises to sequence an entire human genome of 3 billion base pairs in 15 minutes.
Archon X Genomics $10 Million Prize Announcement
The first day's sessions closed with a keynote by Harvard University genetics guru George Church. He forcefully argues that everyone needs whole genome sequencing now! After all, there are 2,700 highly predictive actionable tests implicating 6,000 different gene variants already available.
Church announced during his talk that he and Harvard's Wyss Institute were formally entering the competition for the $10 million Archon Genomics X Prize. The goal is to sequence 100 genomes donated by people who are 100 years old or older between September 5 and October 5, 2013 for less than $10,000 each. The search is on for rare gene variants that protect against disease and enhance longevity. Among many other worthy projects, Church is behind the Personal Genome Project which aims to get 100,000 participants to voluntarily reveal their genetic and medical information so that researchers can begin to link gene combinations more firmly to disease onsets and outcomes.*
Day 2: Friday I will report on the Thursday sessions dealing with designer babies, interpreting the genomic information of healthy people, and what the venture capital world thinks of personal genomics.
*Disclosure: I am a participant in the Personal Genome Project.