Every year brings new smartphones with incremental updates — and the same gnawing problems. Chief among these is their fragile screens. A single fumbled selfie or the keyring crammed into your purse or pocket could be enough to disfigure your pristine, thousand-plus-dollar investment.
When it comes to those responsible for making smartphone glass, there’s a gorilla in the room. Gorilla Glass, a brand of Corning, based in New York, looms large over the market. Steve Jobs himself is why: six months before the launch of the original iPhone, Jobs phoned Corning’s CEO to demand a glass cover with unprecedented hardness. He got it, and now Gorilla Glass can be found on over six billion devices worldwide.
But is glass — even advanced glass — the limit?
For all its luster, glass manufacturing has real drawbacks. Massive polluting furnaces are needed to melt mined minerals into a molten sludge which must then be cooled in controlled settings. Additional complex chemical steps are needed to introduce scratch-resistance and durability. At its heart, this age-old process is clearly bad for the environment.
Arzeda, based in Seattle, thinks there is a better way of making hard surfaces for phones. Their idea starts with the opposite of a blazing furnace — the tulip.
Tulipalin is a natural molecule found in small quantities in tulips that can make bioplastics harder. If it were as easy to brew as beer, bioplastics could become a serious contender for smartphones and other devices. Arzeda has ported the metabolic pathway responsible for making tulipalin into industrial microbes, unlocking a new way to brew up the compound. This work is being pursued in partnership with the Department of Energy.
The key to how they do this is computers. I’ve written how synthetic biologists use computers to understand how genetic sequences encode biological function at the molecular level. Arzeda uses this kind of analysis too, but goes one step further: they have software to model the physics of these systems at the molecular level. Starting with a basic understanding of how different atoms interact with each other, Arzeda can design from scratch the exact proteins it wants.
“Glass is not the only industry process in need of a sustainability upgrade,” says Zangellini.
Last October, Arzeda and BP announced a collaboration aimed at developing a new way of making “a renewable chemical with broad applications.” Though protein design will undoubtedly be key to the process, the companies are staying tight-lipped about exactly what chemical is being pursued.
This secrecy is typical of Arzeda, which makes its money by selling intellectual property — in this case, optimized genetic sequences that encode custom proteins — to other firms. Its customers include Corteva Agriscience (formerly DuPont Pioneer), a major U.S. seed supplier and INVISTA, a major chemical manufacturer.
Beyond what the mind can do
“There are many aspects of protein design that are beyond what a human mind can do,” notes Zangellini. For this reason, proprietary software is king. Arzeda relies on it, along with cloud computing, to find solutions to its customers’ problems.
Once the software cooks up a novel protein sequence, scientists at Arzeda test its performance in lab-grown cells. Advances in DNA synthesis now enable the company to test 10,000 designer proteins per week, Zanghellini tells me.
“We have to test several hundred or several thousands of designs, but it’s a very small number compared to what you would have to screen if you did that randomly, which is more or less what traditional protein engineering does.”
“Machine learning has great potential to find correlations and patterns that are useful for protein design,” Zanghellini continues. But this will depend less on the raw abilities of machine learning algorithms as the quality of the datasets they are trained on.
Zangellini, who holds a master’s degree in computer science, believes his company is uniquely positioned to generate the type of data that will be most amenable to deep learning. If he is right, it will own its own virtuous cycle of improvement; Arzeda-designed proteins would beget data that will reveal better ways of designing proteins. “In the long run, I am absurdly bullish,” he says.
Deep learning is therefore poised to do more than improve the guts of your next smartphone — it might make it scratch-proof, too.
Acknowledgment: Thank you to Ian Haydon for additional research and reporting in this article. I’m the founder of SynBioBeta, and some of the companies that I write about — including Arzeda — are sponsors of the SynBioBeta conference (click here for a full list).
Originally published on Forbes https://www.forbes.com/sites/johncumbers/2019/11/26/molecule-maker-arzeda-wants-to-grow-phone-screens-that-wont-scratch/0