By Andy Fell
March 20, 2019 -- Computer scientists at the University
of California , Davis , and the California Institute of
Technology have created DNA molecules that can self-assemble into patterns
essentially by running their own program. The work is published March 21 in the
journal Nature.
“The ultimate goal is to use computation to grow structures and enable more
sophisticated molecular engineering,” said David Doty, assistant professor of
computer science at UC Davis and co-first author on the paper.
The system is analogous to a computer, but instead of using transistors and
diodes, it uses molecules to represent a six-bit binary number (for example,
011001). The team developed a variety of algorithms that can be computed by the
molecules.
"We were surprised by the versatility of algorithms we were able to
design, despite being limited to six-bit inputs," Doty said. The
researchers were able to design and run 21 algorithms over the course of the
experiments, demonstrating the potential of the system, he said.
Working initially as postdoctoral scholars with Professor Erik Winfree at
Caltech, Doty and co-lead author Damien Woods, now at Maynooth University , Ireland ,
designed a library of short pieces, or tiles, of DNA. Each DNA tile consists of
42 bases (A, C, G or T) arranged in four domains of 10-11 bases. Each domain
can represent a 1 or 0 and can stick to some of the domains on other tiles. No
two tiles are a complete match.
Programmable DNA Lego Bricks
Two of the four domains on each tile are “input,” and two “output.” In an
electronic diode, transistor or logic gate, a value of 0 or 1 at the input (or
inputs) will give a known value at the output. Similarly, depending on which
tiles the researchers selected to begin their program, they could get a known
output at the other end.
Starting with the original six bits of input, the system adds row after row
of molecules, progressively running the algorithm. Instead of electricity
flowing through circuits, rows of DNA strands sticking together perform the
computation.
It’s rather like having a set of Lego bricks, some of which will
spontaneously stick to other bricks. Select a set of bricks to start with, mix
them together and watch them self-assemble into a structure.
The end result of the program is something like a knitted scarf of DNA,
made of tiles stuck together in a pattern set by the original program. The
results are read with an atomic force microscope, which detects a marker
molecule attached to the DNA.
The team was able to demonstrate algorithms for a variety of tasks,
including counting exercises, random walks and drawing patterns such as
zigzags, diamonds and a double helix in the DNA.
Doty and Woods began the work as theoretical computer scientists, so they
had to master some “wet lab” skills. In the future, molecular programming might
operate at a higher level, Winfree said. Today’s coders don’t need to
understand transistor physics, for example.
At UC Davis, Doty is now working on theoretical aspects of molecular
programming. DNA is of special interest because it both represents information
in molecular form, and it is relatively easy to work with, he said.
“It’s a great gift the molecular biologists have given us computer
scientists,” he said.
Additional authors on the paper are Cameron Myhrvold, Joy Hui and Peng
Yin of Harvard University ,
and Felix Zhou of the University
of Oxford . Funding
support came from the National Science Foundation and NASA.
No comments:
Post a Comment