Genepeeks’ Sperm Bank Acquisition Heralds Genome Screening of ‘Virtual Progeny’

By Kevin Davies 

January 4, 2013 | It all began five years ago when Anne Morriss, an economic development consultant, and her partner decided to use a sperm bank to conceive a child.  

They studiously reviewed the profiles of hundreds of potential donors. Morriss searched for signs of empathy in the donor’s essays. “I was obsessed with sentence structure and overuse of adjectives,” she recalls. Her partner, on the other hand, was a frustrated basketball coach and wanted more of a jock, someone who had played sports in high school.  

And there was something else. “It was important for my partner that our children look like me,” says Morriss.  

Anne Morriss

 One donor met all their criteria, but besides physical appearance, genetics was an afterthought. “We trusted that the companies were using bleeding-edge science. It turns out that that was not true,” says Morriss. “We chose a wonderful donor but he was a carrier for a rare Mendelian disorder called MCADD.” Unbeknownst to Morris, she was a carrier too.   

Following recommendations from the American College of Medical Genetics, sperm banks in the United States only screen donors for cystic fibrosis and a couple of other conditions. Most sperm banks will also screen for sickle-cell anemia (in the case of African-Americans) and a panel of so-called Jewish disorders if appropriate. The only other Mendelian disorder screened widely is spinal-bulbar muscular atrophy (SMA).  

“Once in a while, a sperm bank will offer one more test, but it’s very old school way of thinking,” says Morriss. “We’re all walking around with deadly mutations. It’s a conservative space and consumers aren’t yet demanding [more aggressive screening].” 

And although in Morriss’ case, the sperm bank had thoroughly vetted all its donors, there is an incentive for some unscrupulous would-be donors to conceal the truth about their family history.  

MCADD (medium-chain acyl-CoA dehydrogenase deficiency) is a surprisingly common recessively inherited disorder in which an enzyme deficiency renders those affected unable to convert fats to sugar. (The carrier frequency among people of European descent is 1 in 40, although Morriss says there are villages in her family’s native Wales where it’s closer to 1 in 10.) As a result, MCADD patients must eat every few hours or risk serious side-effects, particularly as infants.  

Thankfully for Morriss, MCADD is one of the several dozen disorders tested in the newborn screening panel. Even then, some babies don’t survive in the 72 hours waiting for the results to come back.  

“We were very lucky, we definitely missed a bullet,” says Morriss. “We walk into [Boston] Children’s Hospital and feel he’s the luckiest kid there.” Morriss and her partner had to feed their son every four hours during his first six months of life. But now he can go a full night without food. “We’ve made it through the worst part,” she says. 

Genepeeks Partners 

Lee Silver

As Silver first discussed publicly at CHI’s Consumer Genetics Conference in October, the working model for Genepeeks is to match the sperm bank client with hundreds of potential donors by generating thousands upon thousands of virtual progeny in silico to identify donors who are less likely to be a good genetic match for the particular client. Those individuals can then simply be filtered out of the pool of potential donors reviewed by the client—for example reducing a pool of 400 potential donors to 300—thereby minimizing the chances of an embryo being conceived with a rare genetic disorder. 

Morriss is the first to admit that this sounds rather like a science fiction scenario. “It smacks of designer babies. People winning the narrative are the GATTACA’s,” she says. But she and Silver insist that their technology could dramatically reduce the risks of sperm bank customers conceiving babies with rare genetic diseases. 

“What we can do and maybe nobody else can do yet is anticipate the genetics of reproduction so we can predict the genetic profile of a child,” says Morriss. “The analytic target for us won’t be the parents. It’ll be the child, which allows us to look for autism and diabetes and complex traits.”  

“For what we want to do—ultimately everything—the idea is that this technology will allow us to determine the risk for any condition that has been characterized in human populations. I can’t think of another way to do it for complex diseases,” says Silver. 

Morriss and Silver have brought on a third partner—Jay Remis, a financial services industry veteran now at UBS, with experience in the strategy and growth of young companies. Late last year, Genepeeks raised $3 million in series A venture funding from an investment group including Selway Capital and Columbus Nova.  

Paving the way to launch the screening service, Genepeeks has partnered with a sperm bank—the Manhattan Cryobank, which has about 100 donors. (Both Genepeeks and the Manhattan Cryobank are backed by the same investment consortium.) The companies formalized their partnership on January 1 and plan to co-launch new screening services in the next six months, Morriss says.   

Silver Playbook 

The central Genepeeks plan is to genotype all the donors in the sperm bank as well as its future clients. Silver’s algorithms will then computationally predict up to 1 million theoretical sperm from the donor and analyze the make-up of a potential child from the union of any of those sperm with an egg.  

“The basic idea—the patent is pending—is to take the diploid genomes of two individuals and to create ‘virtual’ gametes,” Silver explains. “It’s a discrete sperm-haploid-genome/egg-haploid-genome coming together to form a discrete diploid genome. That’s a critical piece—you can do probability analysis with carriers, but we can get each discrete [virtual] genome to interrogate the databases. We repeat that process over and over.” 

“It’s not at all a scheme to avoid FDA oversight,” Morriss hastens to add. “It’s a better way to look at reproductive risk.” The screening obviously doesn’t introduce any additional risk to the process, and as she notes, the current sperm bank process is already FDA approved. “We’re just adding another layer and making it safer,” she says.  

“We’re never making a diagnosis of an existing genotype,” stresses Silver. “The diagnosis we make for virtual progeny are for genomes that have never existed—and will never exist.”  

“The virtual genomes don’t provoke anybody—it’s not the genome of the male or female but of an imaginary entity,” he continues. “The sperm bank donors are anonymous in the system. We’re not doing a diagnosis on them. We’re just looking at the virtual entities. Moreover, if they show any risk, the clients never see those donors. The only donors we provide to the clients are the ones that have less risk.” 

Silver will initially work with Affymetrix SNP array data but the method can be extended to exome or whole-genome sequence data. Genepeeks will start by outsourcing the screening of more than 250,000 SNPs, including a considerable number of known disease-causing and nonsense mutations from the 1000 Genomes Project, supplemented with mutations from a range of sources, curated by the company’s scientific team. “We don’t do anything wet—we’re pure information,” says Silver. 

After the customer submits a saliva sample for DNA analysis, the results are compared with those of the selected sperm donors. “Then we do the virtual progeny analysis. Whatever the level of analysis—microarray to WGS—we put her together [with the first donor] and see the risk. We repeat that for each person, run many times.”  

Although the focus will be on Mendelian disorders, Silver says the method could extend to evaluating the ‘virtual’ genetic risk for common diseases such as type 2 diabetes. “If we look at type 2 diabetes, where we have information—not just additive genetics trait—each virtual progeny will have certain risk alleles for interrogation,” says Silver. “You get the risk of diabetes of each [donor], then integrate that information.” 

Silver says that generating phasing information from the genome data is a key step. “There’s a way to represent the phasing in a VCF format—total phasing across the whole genome,” he says. The process of creating virtual haploid sperm is “easiest and most accurate with fully phased genome information.” Silver then uses knowledge of recombination hotspots and frequency to generate chromosomes from male and female using a Monte Carlo probability method.  

“You can’t do this just once—you have to repeat it many times,” he says. “Each time gives us a phenotype. We get 100 or 1000 [sperm] until we reach a point where it’s leveled off [and we compute] a real risk factor. For simple diseases, this doesn’t require that many repetitions. But if you’re including 60 SNPS [for a complex trait], it will be empirical.” 

While Genepeeks has much work ahead to launch its program, Morriss sees a bright future for the technology beyond sperm banks. Down the road, she would like to move into the planned pregnancy space. “I believe anyone thinking about a pregnancy would benefit from this kind of testing,” she says. “The industry is ready to move faster.” 

Morriss recognizes that some critics will ask whether people will want or need this information, but she has no doubts. “Would I rather just roll the dice or find out that the love of my life and I are putting our child at risk if we decide to start a family?”