The grand promise of artificial intelligence has always been the benefit to humanity. We were promised climate change solutions and cures for disease. Instead, it feels like the market is swamped with chatbots and large language models (LLMs) built on scraped and stolen intellectual property, designed to take your job or drive you to delusions. But a London start-up has been diligently working for years to create something that could change the future — for good.
Basecamp Research is the creator of EDEN — a cluster of AI models trained on DNA and evolutionary patterns that can algorithmically deduce new drugs and gene therapies to treat cancers and the rarest of genetic diseases. It is trained on BaseData, the company’s genomics dataset. And it’s backed by the computing power of Nvidia, the tech giant powering the latest AI innovations.
“EDEN is one of the largest AI models ever trained,” Basecamp co-founder Oliver Vince tells me. “It’s the largest biological model ever trained. It’s the thing that understands biology the best. We’re trying to learn from nature to design solutions for the future of the planet and humans.”
Right now, they have a lab in Boston demonstrating how the model can develop treatments for cancer and genetic disease. In January, they announced they’ve designed a therapeutic enzyme that can insert large amounts of DNA at a precise point in the human genome to treat a specific disease. Think CRISPR, but with larger genome edits and without DNA damage.
Via the enzyme, DNA could be inserted into the immune system to produce cancer-fighting cells, or to repair faulty genes to cure a sick baby. On a grander scale, EDEN could be the key to vanquishing superbugs — the antibiotic-resistant pathogens that keep doctors up at night — by developing novel antimicrobial peptides.
“All of this sounds very sci-fi,” admits Vince. “And we’re not there yet. But our ultimate goal is that this is a tool that all doctors use.” Currently, developing a new drug is a time, labour and resource-intensive process that can take years, costing millions of pounds. And patients with rarely seen genetic disorders often don’t have that kind of time.
“The only way to design something that complex is going to be with a computer. And the only way to get a computer to understand biology is to go out and hunt for it”
With EDEN, medical researchers can input a patient and disease profile and have the AI model reason out a targeted cure. All this from DNA samples retrieved from deep-sea volcanoes and shipwrecks, remote rainforests and rubbish dumps.
Here be (bio) pirates
Basecamp Research is built on fieldwork. Vince was working on treating brain cancer for his PhD at Oxford, while his co-founder Glen Gowers was doing his own PhD in synthetic biology at Imperial College, finding ways to design DNA to make new HIV drugs. They dreamt of a more efficient way to design more sophisticated medicines. They had a realisation. “The only way to design something that complex is going to be with a computer,” says Vince. “And the only way to get a computer to understand biology is to go out and hunt for it.”
In 2019, they set up a base camp on an icecap in Iceland with their mobile DNA-sequencing lab. Solar-powered and able to fit on a sledge, it was proof of concept. “We were in a tent, running this miniature lab in the middle of nowhere,” says Vince. “It gave us the spark of what we could actually go and do.”
EDEN is so large because it has been trained on 10 trillion tokens of evolutionary DNA from a million species. For five years, Basecamp has been collecting it in the field from 150 locations in 28 countries across five continents. This is bioprospecting: collecting biological data from indigenous species from different countries and using it to develop socially – and commercially – valuable materials, including pharmaceuticals.
But bioprospecting has a dark side: biopiracy. Indigenous communities in less developed nations may have centuries of knowledge, only for a large corporation to commandeer the biological data, slap a patent on it and sell it for profit they won’t see a penny of. Basecamp has taken great pains to avoid this kind of thievery.
.jpg)
According to the Nagoya Protocol, every country has the right to their genetic resources and associated traditional knowledge. Therefore, an Access and Benefit-Sharing scheme should be devised to ensure an equitable distribution of the benefits created from outside parties gaining access. “We’ve pioneered a way of doing that for training AI systems,” says Vince. “We have built partnerships with countries and the people on the ground who are experts in the local environment, and they get investment, resources and training.
The ultimate goal is a kind of circular economy where everyone wins. “On one side, you've got people making medicine, and hopefully curing people. And to feed that, you've got lots of people working hard to study nature and to feed it into the ultimate medicine-making machine.”
That “medicine-making machine” has its own strict guardrails. After all, such a vast trove of DNA could also be used to develop all kinds of novel poisons and bio-weapons. It’s not able to dream up a new infectious disease, Vince assures me, and it’s not going to be available to the general public to cook up their own drugs.
“This is not a free-for-all toy,” he explains. “It's being built for doctors to use, and you give doctors chemicals that also can be harmful. You are giving a tool of great responsibility to people who have great responsibility.”
A genetic atlas of the world
The faster these new drugs can be designed, the more patients suffering can be alleviated. So the next step is to bank even more genomic data to power up the models. In March, Basecamp launched the Trillion Gene Atlas, in partnership with Anthropic, Ultima Genomics and PacBio.
The goal is to map a trillion genes, sourced from the genomic data of 100 million species of plants and animals. “It’s a massive expansion of what we know about life on Earth,” enthuses Vince. “It’s one of the largest sequencing projects the world has ever done. It will be one of the biggest computer projects ever done.”
It’s also potentially a watershed moment for tech companies applying their resources to a project of huge social value. It demonstrates the everybody-wins, collaborative mentality of good science. If all the silicon chips and water-hungry data centres are going to take from the Earth, you should be able to benefit humanity, Vince explains.
“If you're trying to make a chatbot slightly more funny, or sycophantic, it's the very competitive nature that is destructive,” he says. “These big companies are falling over themselves to compete, whereas in science, there are so many big problems people are building on each other and building on success.”
“The UK has such great scientific talent and depth, and you need to combine that with AI to make progress”
Vince also hopes it can demonstrate the huge, currently rather untapped potential of UK innovation. “This is where the UK is going to prosper in AI, not specifically this, but in the sciences plus AI,” he says. “The UK has such great scientific talent and depth, and you need to combine that with AI to make progress.”
To make that happen, Vince thinks there needs to be more excitement ginned up this side of the pond. He’s noticed a distinct divide, where Americans are willing to risk failure, whereas the British focus too much on the potential to fail.
“We want people to say, we’re going to take some scientific leaps in this country. Some of them are not going to work — but some are going to change the world.”
