Lux Research recently attended the EmTech Digital Conference in San Francisco. At the event, Deep Genomics CEO Brendan Frey described how the company wants to use a form of artificial intelligence (AI) called neural networks to connect genotype to phenotype – in other words, to see how the genetic code of an individual determines the traits (hair color, disease, persona) the individual presents. To illustrate the gap, he surveyed the audience by show of hands, how many believed that understanding genomics would change our lives (some 80%); how many had actually had a genetic test done (perhaps 25%), and how many of those had found the results to be valuable (about 5%). “That’s the gap,” he explained, since so many conditions lack genetic correlation, and so many genetic traits are not actually expressed. The goal of Deep Genomics is to complement genomic data with other biologically-relevant data, including personal history, behavior, environment, and nutrition; and use machine learning to connect genes with traits. As a first proof of concept, the company has released a database and Science paper describing 328 million single nucleotide variants (SNVs) – genetic mutations – and their identified effects on RNA splicing (a biochemical process that happens as DNA is translated into proteins).
Obviously, Deep Genomics’ immediate payoff lies in its ability to predict disease and appropriate interventions on a personal level. But it also foretells the future of computing and analytics, as biological processes inspire and inform biomimetic computing approaches that we have only begun to understand today. The now-rudimentary understanding we gained in the 1940s of how the brain functions led to the 1980s theoretical, and now practical, underpinnings of the AI class of artificial neural networks, and companies like Nervana Systems. Precision medicine will also help us learn from the immune system’s amazing ability to fend off novel attacks, and apply it to cyberphysical security (see the report “Cybersecurity Venture Investment in Pervasive Computing and the IoT“); genetic regulatory processes will help us design better mechanisms for managing cities and societies, and what we learn about organisms’ ability to maintain biostasis in turbulent environments will inform corporate strategy.
Mark Bünger is a Vice President of Research at Lux Research an co-chair of AI World Conference & Expo