Scientists have been experimenting with ways to use DNA as a data storage medium, but it’s difficult to retrieve and manipulate data written to it. Now a team has developed “chemical neurons” that can conduct calculations on data stored in DNA and read back the answers easily.
Our modern data storage systems may be impressive, but as with many things nature has done it much more efficiently than anything we've achieved. A single gram of DNA can store up to 215 million GB of data, which theoretically means the contents of the entire internet could be stored in something the size of a shoebox. Better yet, under the right conditions DNA could last thousands or even millions of years.
It’s not surprising then that scientists have been investigating ways to tap into that. Books and albums have been stored on DNA, while others are exploring how to improve data density, stability, and the ease of writing to or reading from DNA.
In the new study, researchers from the CNRS, the ESPCI Paris-PSL and the University of Tokyo focused on improving ways to find specific pieces of data stored on DNA, and performing calculations with it.
The team created chemical neurons using three enzymes that have specific reactions with each other, allowing them to essentially pass information around like natural neurons in the brain. These neurons were then structured into an architecture combining multiple layers that function like a neural network.
These chemical neurons are able to perform calculations on data stored in droplets containing DNA, and will express the results by emitting fluorescent signals that can be interpreted by other instruments. Using microfluidic systems to shrink down the enzyme reactions, the team says that tens of thousands of these reactions can take place.
This technique could not only help scientists find and process data contained inside huge DNA databases, but they could also eventually help detect biomarkers of disease in blood tests or other liquid biopsies.
The research was published in the journal Nature.
Source: CNRS
