Jun Wang, who stepped down as CEO of Chinese sequencing giant BGI last month, has told Nature News what he wants to do next. The tentative plan is to raise $1.6 billion, gather samples from 1 million people and analyze them with a supercomputer to uncover a "formula for life."
Many details of the plan are yet to be finalized. Wang is thinking of raising $1.6 billion to finance the construction of a 1 million-sample prototype of the platform--eventually he would like 100 million people in the database--but is yet to establish a business model or start trying to persuade investors to part with their money. The elevator pitch is ready, though. Having had a hand in many of the big genomic programs of the past decade, Wang now wants to improve human health by creating a "million-digitized-lives project" that will define the next 20 years of his career.
"The end goal is to develop an ecosystem. It will be a virtual village. When people 'stay' in the virtual village, it will advise them on how to live more healthily and for longer, including hints about what they should eat and what exercise they should be taking. All the advice will be based on various factors including genetic make-up and lifestyle. Individuals, doctors, researchers, pharma companies will all be part of it," Wang said. The project will act as an ideas factory for drug developers. "If the AI system can't find the answers, maybe the pharma companies will. The data will still be valuable."
Such a vision has echoes of Lee Hood's Arivale, J. Craig Venter's Human Longevity Inc., and Google's ($GOOG) Calico. Wang, in keeping with the spirit of his work at BGI, is likely to compete well in terms of sheer scale. The ex-BGI CEO already has 100,000 samples and thinks he needs 1 million to make the project work. Beyond this, Wang wants to keep growing, up to 100 million samples if possible. The genomic data contained in these samples are central to the initiative but are only one component. Wang also wants to gather a total of one terabyte per person of omics data and other information.
Algorithms running on supercomputers will then look for links between a person's genes, lifestyle and environmental data points to predict their phenotype." The AI will try to draw a formula for life. Life is digital, like a computer program--if you want to understand the results of the programming, how the genes lead to phenotypes, it is sufficiently complicated for you to need an AI system to figure out the rules," Wang said. "People have told me I had crazy ideas before. But you know, it turned out pretty good. I am a risk-taker. I am betting all of my credibility on this."
- read the interview