We’ve seen AI algorithms create mostly-believable images of people’s headshots, works of art, news stories, and even pokémon. But now a team of scientists is taking this a step further, with an algorithm that produces the whole genetic code of non-existent humans.
A versatile collection of molecular methods allows synthetic biologists to assemble DNA of various sizes, from genes to chromosome scales. Synthetic genome is a synthetic genome whose creation requires either genetic alteration of pre-existing life forms or artificial gene synthesis to generate new DNA or whole life forms. The discipline used to study synthetic genomes is called Synthetic Genomics.
A team of scientists is taking things a step further, with an algorithm to generate the entire genetic code of nonexistent people.
Using a form of AI called the Generative Adversarial Network (GAN), in which two algorithms quickly produce some kind of output, review their work against real-world samples, and refine things as they go, the team has managed to reproduce realistic human genomes that they believe are indistinguishable from the real thing, according to the research first spotted by The Next Site.
Unlike other GAN projects—like those terrifying pokémon—which are supposed to be fun little AI oddities, the team of researchers from Estonia’s University of Tartu and France’s Paris-Saclay University who have created an artificial genome project say that their fake genetic sequences have real utility as a tool for research geneticists.
They claim in their article, which was published Thursday in the journal PLOS Genetics, that these DNA codes may support further genetic experimentation—without violating the privacy of real individuals who would have to give up their genetic data.
But it may not be what it’s cracked to be, the unaffiliated geneticist told Futurism. Consider the other explanations of what a GAN will build to understand why. We know what the human form, the popular pieces of art, and the pokemon look like. So, for AI, learning from these examples and constructing one’s own is no challenge. Although as far as the human genome is concerned, there are already so many concerns about what the various sections do, and they mean that it is not even possible to plug it into an algorithm as training data.
“My initial opinion is that this is interesting, but I’m not sure that I can see any real practical implications for research right now,” said Deanna Church, vice president of the Mammalian Business Field and Software Policy at Inscripta biotech corporation, Futurism. “A lot of the scientific community is still trying to figure out how the non-coding (most) of the genome ‘works.’ Just because you can’t computationally distinguish these generated genomes from real genomes doesn’t mean that they’ve actually preserved important functional patterns and domains—a there’s lot of that we still don’t understand.”
The real question in our minds is whether these GAN-generated genomes will eventually compile, so to speak, into a real, usable human genetic sequence. The authors of the paper did not respond to Futurism’s call for feedback. But in that sense, Church says, she’s not too confident.
“As far as ‘compiling’ this into a human—I certainly don’t see the way to that,” Church said to Futurism. “There’s a long way from computationally generating genome sequences to converting them to human cells, much less human ones.”
“I’m sure the work is interesting, but I don’t see any practical applications for it right now,” Church said. “Of course, there’s something I could miss.”
All species on our world store the genetic blueprint of life in the DNA code of their genome. The computational biology revolution, powered by DNA sequences, enables us to read the genomes of countless microbes and multicellular species that inhabit our world. Currently, DNA sequences of more than 200,000 microbial genomes are preserved in digital genome libraries and have massively improved our understanding of how DNA programs live structures. Using this amazing treasure chest of molecular building blocks, bio-engineers learn how to read (sequence) and write (using chemical synthesis) long DNA molecules and how to produce useful microbes with the aid of computers.