A DNA computer completed 100-bit calculations by relaxing toward equilibrium

The discovery

The researchers asked whether a programmable physical computer could be designed so that the correct answer is the most thermodynamically favored state, allowing molecular components to settle toward the answer instead of relying on tightly timed reaction cascades and separate error-correction machinery.

The research question and why it matters

DNA computing dates to demonstrations that molecular reactions could solve combinatorial problems, followed by strand-displacement circuits, self-assembling algorithms, molecular robots and DNA systems that combine data storage with computation. Most operate away from equilibrium and manage unwanted reactions through sequence design, timing or redundancy. This work instead makes the intended final arrangement the favored equilibrium state and tests that principle across several programs, speeds and scaffold sizes.

What researchers found

The platform executed 10 programs, including addition, multiplication by three, division by two, parity, balanced-bracket recognition, counting, graph reachability and a small Rule 110 calculation. Four-position programs averaged 95.3% estimated yield under the usual protocol; addition averaged 96.7%. Fast runs produced separable outputs in 30 seconds to one minute, with 81.2% average estimated yield. A 25-position system added two 25-bit numbers in roughly 14 hours, representing 100 bits when inputs, carry states and outputs are counted. A simple bit-copy program was renewed 25 times.

Results at a glance

Key results from the tested systems

>700

computations

Experiments covered 10 distinct molecular programs.

100 bits

largest demonstration

Two 25-bit inputs, 25 carry bits and 25 output bits were encoded on a 25-position scaffold.

95.3%

mean estimated yield

Reported across the short four-position programs under the usual annealing protocol.

30–60 sec

fastest runs

Short programs produced separable fluorescent outputs in under a minute.

How the research worked

A single DNA scaffold was divided into addressable positions. At each position, competing DNA compute strands encoded program instructions and input bits. Correct neighboring strands formed energetically favored matches, while mismatches could be replaced during controlled heating and cooling. Fluorescent reporters read selected output positions. The team compiled finite-state programs into strand sets, tested ordinary three-hour and fast annealing schedules, renewed inputs by adding blocker and replacement strands, and expanded the scaffold to 25 positions.

Subjects or systemLaboratory

Research designExperimental DNA nanotechnology study combining thermodynamic design, fluorescence readout, mathematical analysis and more than 700 molecular computations

Evidence baseTen DNA programs and more than 700 computations, ranging from short four-position systems to a 25-position scaffold that encoded two 25-bit inputs, 25 carry bits and 25 output bits—100 bits of computation

How to interpret this design

A controlled experiment can isolate a mechanism under defined conditions. The tradeoff is external validity: performance in a laboratory system may change when materials, organisms, environments or operating constraints differ.

The evidence comes from a controlled physical or chemical system. That control helps establish what happened under the tested conditions, while scale-up, durability, manufacturing and real-world performance remain separate questions.

What strengthens or limits the finding?

The team tested 10 programs in more than 700 computations, reported quantitative yields, demonstrated reuse and scale-up, and released source data and code. Performance declined in the largest systems, readout required bulk laboratory instruments and the study did not compare useful work per joule with conventional computers.

The result is meaningfully informative, but identifiable limitations could alter the size, reach or causal interpretation of the finding.

Funding and disclosure context

The recorded funding source is: European Research Council Active-DNA grant 772766; European Innovation Council DISCO grant 101115422; Science Foundation Ireland grants 18/ERCS/5746 and 20/FFP-P/8843; and Research Ireland grant 24/PATH-S/12367. The recorded conflict information is: Tristan Stérin, Abeer Eshra and Damien Woods are inventors on pending Maynooth University patent applications EP4715680A2 and US20260065010A1 covering core principles and fluorescent reporting; Stérin worked for PRGM DEV during the latter part of the project. Funding or a disclosed relationship does not by itself invalidate a result, but it is relevant when judging design choices, analysis and the need for independent replication.

What it means

The experiment shows that useful logic can be embedded in an energy landscape: molecular components can correct many local mismatches as the mixture approaches equilibrium. That could inform specialized biochemical computing, sensing or materials that process information where DNA chemistry is useful. The main advance is an experimentally programmable architecture, not evidence that molecular computers are ready to outperform electronic hardware.

Beyond the abstract

Deeper analysis

The answer is built into the energy landscape

Each program chooses strands whose correct neighborhood has fewer energetic penalties than competing arrangements. Heating allows strands to exchange; cooling makes the lowest-energy configuration increasingly favorable. Computation is therefore the physical relaxation process itself.

Programmability distinguishes the result

The same scaffold design can run different finite-state algorithms by changing the strand set. Demonstrating 10 programs matters because a one-off molecular calculation could reflect a custom chemical trick rather than a reusable computing model.

Speed and scale pull in opposite directions

Small systems produced readable answers rapidly, while the largest needed long anneals and delivered lower yield. That tradeoff is the central engineering challenge and a better guide to the result than the headline bit count alone.

Efficiency remains unmeasured

A thermodynamically favored final state reduces the need for continuous error-suppression reactions inside the molecular system. But system-level efficiency cannot be claimed without counting strand manufacture, temperature control, liquid handling and optical readout.

Keep the claim in proportion

What it does NOT prove

It does not show that the DNA system is faster, cheaper or more energy-efficient than a silicon computer for practical tasks.
It does not create a general-purpose 100-bit processor with memory, an operating system or arbitrary software.
It does not demonstrate computation inside living cells or a medical diagnostic or treatment.
It does not eliminate energy use: DNA synthesis, heating and cooling, liquid handling and fluorescence measurement all require resources.
It does not show error-free scaling to thousands or millions of molecular positions.

Important limitations

Most detailed performance measurements used four-position scaffolds and pairs of technical repeats, far smaller than practical digital workloads.
The 25-position calculations required about 14 hours rather than the seconds achieved by short programs.
Estimated yield fell during scale-up: the reported bit-copy yield was 71% at 20 positions and 59% at 25 positions.
Bulk fluorescence reports an ensemble average and does not directly read every individual molecular assembly.
The 100-bit description counts input, intermediate carry and output information; it is not equivalent to a conventional 100-bit CPU or 100 bits of addressable memory.
Patent interests and one author's commercial employment make independent replication particularly important.

How this fits with previous research

Questions still unanswered

Can the architecture scale while maintaining high yield and reasonable computation time?
What is the complete energy and material cost when strand synthesis, thermal cycling and readout are included?
Can individual molecular outputs be read reliably without bulky fluorescence equipment?
Which sensing, storage or biochemical-control tasks benefit enough from molecular operation to justify the slower speed?
Will independent laboratories reproduce the reported yields, renewability and scale-up behavior?

Government verification and context

Relevant U.S. government resources

These resources serve different purposes. A registry can verify what researchers planned, a repository can locate government-funded work, and an agency page can supply authoritative background. None automatically proves that this paper's conclusion is correct.

Government repositoryU.S. Department of Energy, Office of Scientific and Technical Information

OSTI.GOV research search ↗

DOE's research repository is used to locate related national-laboratory reports, accepted manuscripts and funding-linked technical work.

Reuse note: Facts and discoveries are summarized here in original language. We link to government material instead of copying it wholesale, and we do not reuse agency logos, photographs, charts or third-party material unless the specific reuse rights are verified.

Sources and provenance