“You’re a male, you’re 34-years-old and your 5-foot-10 ½ inches tall.”
That’s a prediction that Craig Venter, long a pioneer on the frontier of genomics, offered to me while we spoke on the phone. He took a wild guess after five minutes of conversation, to illustrate the kinds of things his researchers at Human Longevity Inc. are now predicting thanks to genetic data. Had he recorded the conversation and fed it through his computers, he says he could have nailed my precise age (32) and height (6-4).
Venter is leading efforts to use digital technology to analyze humans in ways we never have before, and the results will have huge implications for society. The latest findings he described are currently being written up for scientific publications. Venter didn’t want to usurp the publications, so he wouldn’t dive into extensive detail of how his team has made these breakthroughs. But what he did share offers an exciting and concerning overview of what lies ahead for humanity. There are social, legal and ethical implications to start considering. Here are five examples of how digitizing DNA will change the human experience:
1. Crime solving
Venter describes the forensic implications of DNA as “huge,” and said he’ll likely develop a related product because of the opportunity. Venter’s team at Human Longevity has been able to predict the faces of individuals from only their DNA. If suspects leave their DNA at a crime scene, they might as well be leaving a picture of their face.
So how is that even possible? Venter has built a huge database of genetic information, plus related body measurements. With 3D photographs captured in his lab, a human face is broken into 30,000 unique measurements. Computers running sophisticated programs then crunch this information and recognize patterns that a human would never see. As the database grows bigger the machines will have more examples to learn from, and their predictions should grow even better. Venter has 20,000 genomes right now, but speaks of over a million by 2020.
While the facial predictions are currently somewhat blurry, Venter expects that to improve. That’s reasonable to expect given the steady advances we’re seeing in computing power and software.
Venter described taking the facial images produced from DNA, and cross-checking them against every photo on the Internet. This could quicken the pace of tracking down a suspect.
“It gives a different use for [sites like] Facebook than they probably had in mind in the beginning,” Venter said. While suggesting potential benefits for law enforcement and national security, he warned of privacy implications and the negative implications for closed dictatorships. All of these new services could be used for good or bad.
2. Predict what your child will look like
Venter said there’s discussion of screening fetal cells from the maternal bloodstream to predict what the child would look like at age 18. But he said the team is concerned about unmasking non-paternity, in which the biological father is not who it is believed to be. He said he wouldn’t want to be the one breaking that news to a couple.
3. Find missing persons
If a person goes missing as a child, and an active search continues for years, no current photos will be available, hampering the search. But with images projected from DNA, better photos could be generated.
4. Predict age and height from a voice recording
In the future someone may only need to have a brief recording of you speaking to pinpoint these traits. This ability might be of interest to college sports recruiters or singles on dating Web sites, as individuals have incentives to describe their attributes in the most favorable light.
Venter says his group has been able to predict a subject’s gender, age and height from a voice recording. They also saw a correlation between voice and face shape. They did that by having 1,000 subjects speak for a few minutes. They also took extensive body measurements — such as throat and jaw size — with MRI machines. Then the team sequenced the subjects’ genomes. All this information gave the machines a chance to learn and make predictions. Venter said the ability to predict height was a surprise.
What’s potent here is turning biological information into digital data so that cutting-edge software can be applied to it and decipher more than what a human could do on their own. The machine learning team at Human Longevity is led by Franz Och, who previously built Google’s Translate software. They view the genome as the software of life that codes us all, and are trying to predict as much as possible from the genome.
5. Unveil secrets we’d prefer were private
Venter, who has his sequenced genome posted on the Internet, said he might rethink that decision if he were redoing the process today, given the rapid growth in what can already be predicted.
The work has shifted how Venter views a classic debate in psychology — nature vs. nurture. Is our environment or heredity more influential?
“I probably started out in the 50-50 camp when I didn’t know anything,” Venter said, who has since shifted largely toward nature. “Your DNA software encodes for who you are, but it also codes for the variability and the plasticity. So a key part of human development is how adaptable we are. Those are parameters that are also built into our genetic codes.”
If you’re interested in learning more about Venter’s work, here is video of a speech he gave at the Council of Scientific Society Presidents: