Biocomputing

Science Fiction Coming to Life

As our world becomes increasingly digital, our energy consumption keeps climbing. Quantum computing might help us keep up with growing computational needs, but it won’t solve the energy problem. Add to that the immense challenge of storing all the data we generate every day, and it’s clear: we need new ideas.

Help might come from a very unexpected direction: biology.

At a second glance, biology and computer science actually share some overlap. Neuronal networks, for instance, are natural computers. DNA, on the other hand, is an incredibly efficient storage medium. So why not use biology to balance out some of the weaknesses of silicon-based computing?

Why biology?

Neuronal networks and brains are exceptionally energy-efficient. For the same computational task, a human brain outperforms the latest supercomputer by at least a factor of 10⁶ in energy efficiency.
And while AI models need massive training datasets to learn complex patterns, the human brain fills gaps quickly, using a million times fewer examples. Even small brains, like those of insects, can achieve similar learning outcomes with far less data.

(Source: Frontiers in Science, 2023)

That said, recreating this efficiency in a lab dish is another story. In vitro systems (cell culture outside of a body) still rely on energy-intensive cell culture methods, so the real energy efficiency of biocomputers remains to be seen.

In short, scientists hope that biocomputing can reduce energy consumption while making computing more flexible and adaptive, especially for solving complex problems with lots of unknowns.

How does that actually work?

Biological systems react to changes in their environment and send out analog signals. That means they can perform logical operations like if A and B, then output signal.

For example, when triggered by their environment neurons will communicate via electrical impulses. But also less complex biological systems consisting only of molecules rather than whole cells can send out signals, for example when molecules are labeled with fluorescent dyes, an optical signal can be released through a reaction triggered by a change in the environment. Both signal types can be picked up by sensors, converted into digital information, and processed by a computer.

The feedback loop works the other way too. A computer can send digital signals back, translated into physical changes through temperature shifts, molecular releases, or pH adjustments.

If the biological system is a neuronal network with enough maturity, it can actually learn which actions lead to more favorable environmental conditions and avoid those that don’t. Over time, the neurons “prefer” the pathways that lead to the pleasant outcomes.

Using this principle, scientists have taught a small group of neurons (fewer than a cockroach’s brain) in a dish to play a simplified version of the arcade game Pong. Through electric signals, the cells were informed about the position of the ball. The at first random signals that the cells send out in return were used to move the paddle in order to connect the neurons to the game. Only when the paddle hit the ball the neurons received a pleasant electrical stimulation from the researchers.

An undoubtedly very impressive work and fun real-life application, but “playing” with neurons is also a very tedious work and far from solving complex mathematical problems. The next generation of biological intelligence involves the use of brain organoids, which are small agglomerates of stem cells that develop into brain cells and self-organize into structures similar to the brain during early embryonic development. Scientists at Johns Hopkins University are aiming to develop a biocomputer based on organoid intelligence (OI) with the aim of outperforming silicon-based computers.

Besides neuronal networks, fungal networks, a.k.a. mycelium, have been used by scientists for biocomputing tasks.

DNA as data storage

A simpler biocomputing solution, which is closer to commercialization, is storing data in the form of DNA.

While digital data uses the binary system (0s and 1s), DNA encodes information in nature through genes. The information encoding parts of genes are the four bases adenine (A), thymine (T), guanine (G), and cytosine (C). Scientists have developed different code formats and methodologies over the years for translating 0 and 1 -based  data to A,T,G, and C -based data in the form of synthetic DNA. (This type of DNA does not contain any genes with information for living organisms.)

Data written into DNA can be preserved for thousands of years if stored properly, conventionally in a stabilizing liquid under frozen conditions. Researchers have already successfully encoded entire books, music albums, movie clips, or Wikipedia into DNA and retrieved them without error.

However, the process is still too costly for commercial use and the synthesis is not yet energy efficient. As synthetic biology advances, though, costs are expected to drop significantly. When that happens, DNA data storage might become a real alternative.

Limitations and outlook

Can biocomputing ever compete with quantum computing? Probably not anytime soon.

Current innovators in the field like to reference the astonishing efficiency and computational power of the human brain. But let’s be realistic, brain organoids or neural cultures in the lab are far from reaching the size or complexity of even a mouse brain. It’s more accurate to compare today’s systems to something like the brain of an Etruscan shrew.

Whether the technology will ever reach the capabilities of a rat or even a dog brain remains to be seen. Even though startups like FinalSparks and Cortical Labs have already shown early applications, proving that connecting neurons to digital systems isn’t just a sci-fi dream anymore.

The takeaway

Biocomputing won’t replace quantum or traditional computing anytime soon. But it might complement them through bridging biology and computer technology in ways that make our digital world more flexible, adaptive, and sustainable.

Even if neurons playing Pong still sounds like the dream of a mad scientist, the fact that we can do it says something profound:
We have reached an understanding of biology and computer science that allows us to connect both worlds. How fast this technology can mature depends highly on the advancements made in synthetic biology and on how quickly scientists can overcome current biological limitations in cell culture. At the present moment, biocomputing will demand a massive amount of research to reach a stage where it can be a relevant alternative to silicon-based systems. Only if you -as an investor- are prepared for a very long development process and want to be involved from the stage of infancy, investing in biocomputing might be interesting.

If this topic sparked your interest, I´d be very happy to talk more!