Teaching a machine to spot a crystal

June 21, 2018, Duke University
Not all protein crystals exhibit the colorful iridescence of these crystals grown in space. But no matter their looks, all are important to scientists. Credit: NASA Marshall Space Flight Center (NASA-MSFC)

Protein crystals don't usually display the glitz and glam of gemstones. But no matter their looks, each and every one is precious to scientists.

Patrick Charbonneau, a professor of chemistry and physics at Duke, along with a worldwide group of scientists, teamed up with researchers at Google Brain to use state-of-the-art machine learning algorithms to spot these rare and valuable . Their work could accelerate drug discovery by making it easier for researchers to map the structures of proteins.

"Every time you miss a crystal, because they are so rare, you risk missing on an important biomedical discovery," Charbonneau said.

Knowing the structure of proteins is key to understanding their function and possibly designing drugs that work with their specific shapes. But the traditional approach to determining these structures, called X-ray crystallography, requires that proteins be crystallized.

Crystallizing proteins is hard—really hard. Unlike the simple atoms and molecules that make up common crystals like salt and sugar, these big, bulky molecules, which can contain tens of thousands of atoms each, struggle to arrange themselves into the ordered arrays that form the basis of crystals.

"What allows an object like a protein to self-assemble into something like a crystal is a bit like magic," Charbonneau said.

Even after decades of practice, scientists have to rely in part on trial and error to obtain . After isolating a protein, they mix it with hundreds of different types of liquid solutions, hoping to find the right recipe that coaxes them to crystallize. They then look at droplets of each mixture under a microscope, hoping to spot the smallest speck of a growing crystal.

"You have to manually say, there is a crystal there, there is none there, there is one there, and usually it is none, none, none," Charbonneau said. "Not only is it expensive to pay people to do this, but also people fail. They get tired and they get sloppy, and it detracts from their other work."

The machine learning software searches for points and edges (left) to identify crystals in images of droplets of solution. It can also identify when non-crystalline solids have formed (middle) and when no solids have formed (right). Credit: Duke Research Blog

Charbonneau thought perhaps deep learning software, which is now capable of recognizing individual faces in photographs even when they are blurry or caught from the side, should also be able to identify the points and edges that make up a crystal in solution.

Scientists from both academia and industry came together to collect half a million images of protein crystallization experiments into a database called MARCO. The data specify which of these protein cocktails led to crystallization, based on human evaluation.

The team then worked with a group led by Vincent Vanhoucke from Google Brain to apply the latest in artificial intelligence to help identify crystals in the images.

After "training" the deep learning software on a subset of the data, they unleashed it on the full database. The A.I. was able to accurately identify crystals about 95 percent of the time. Estimates show that humans spot crystals correctly only 85 percent of the time.

"And it does remarkably better than humans," Charbonneau said. "We were a little surprised because most A.I. algorithms are made to recognize cats or dogs, not necessarily geometrical features like the edge of a crystal."

Other teams of researchers have already asked to use the A.I. model and the MARCO dataset to train their own machine learning algorithms to recognize crystals in protein crystallization experiments, Charbonneau said. These advances should allow researchers to focus more time on biomedical discoveries instead of squinting at samples.

Charbonneau plans to use the data to understand how exactly proteins self-assemble into crystals, so that researchers rely less on chance to get this "magic" to happen.

"We are trying to use this data to see if we can get more insight into the physical chemistry of self-assembly of proteins," Charbonneau said.

Explore further: Machine learning offers new way of designing chiral crystals

More information: Andrew E. Bruno et al. Classification of crystallization outcomes using deep convolutional neural networks, PLOS ONE (2018). DOI: 10.1371/journal.pone.0198883

Related Stories

Machine learning offers new way of designing chiral crystals

April 10, 2018

Engineers and chemists at Hiroshima University successfully used the technology underlying facial recognition to design chiral crystals. This is the first study reporting the use of this technology, called logistic regression ...

Microseeding: A new way to overcome hemihedral twinning?

December 15, 2016

Twinning is a crystal-growth disorder in which the specimen is composed of distinct domains whose orientations differ but are related in a particular, well-defined way. Twinning, which is a known problem in protein crystallography, ...

Space station crew cultivates crystals for drug development

March 31, 2017

Crew members aboard the International Space Station will begin conducting research this week to improve the way we grow crystals on Earth. The information gained from the experiments could speed up the process for drug development, ...

Physicists find signs of a time crystal

May 2, 2018

Yale physicists have uncovered hints of a time crystal—a form of matter that "ticks" when exposed to an electromagnetic pulse—in the last place they expected: a crystal you might find in a child's toy.

Cells target giant protein crystals for degradation

March 12, 2015

Researchers at the RIKEN Brain Science Institute in Japan engineered a fluorescent protein that rapidly assembles into large crystals inside living cells, and showed that cells actively targeted the crystals for degradation. ...

Molecular details of protein crystal nucleation uncovered

April 4, 2018

A team of researchers led by Dr. Mike Sleutel from the VIB-VUB Center for Structural Biology in collaboration with scientists from the Institute for Complex Molecular Systems of the Eindhoven University of Technology, and ...

Recommended for you

Reducing the impact forces of water entry

November 20, 2018

When professional divers jump from a springboard, their hands are perpendicular to the water, with wrists pointed upward, as they continue toward their plunge at 30 mph.

Tiny lasers light up immune cells

November 20, 2018

A team of researchers from the School of Physics at the University of St Andrews have developed tiny lasers that could revolutionise our understanding and treatment of many diseases, including cancer.

1 comment

Adjust slider to filter visible comments by rank

Display comments: newest first

knowphiself
not rated yet Jun 21, 2018
I'd put my money on electron holography

Please sign in to add a comment. Registration is free, and takes less than a minute. Read more

Click here to reset your password.
Sign in to get notified via email when new comments are made.