Imagine storing poems, photographs, videos or other digital information inside molecules of DNA—and then retrieving that information without needing thousands of copies of each DNA fragment.
Scientists led by Weigang Chen have developed a new DNA data-storage method designed to make this possible. The approach combines medium-length DNA molecules, low-density parity-check (LDPC) codes and pseudo-noise sequences to overcome one of the biggest problems in DNA storage: errors during nanopore sequencing.
The researchers demonstrated that digital data could be recovered reliably from DNA using sequencing coverage as low as 1.24–3.15×, with typical nanopore error rates around 1.83%. In particularly favorable cases, recovery was possible at approximately 1× coverage, approaching a remarkable single-molecule readout scenario.
Why Store Data in DNA?
As the amount of digital information produced worldwide continues to grow, conventional storage technologies face challenges involving physical space, energy consumption and long-term preservation.
DNA offers a very different possibility. A tiny amount of DNA can theoretically hold enormous quantities of information. It is also chemically stable when properly preserved and requires very little energy to maintain compared with continuously powered electronic storage.
This makes synthetic DNA particularly attractive for cold data storage—information that needs to be preserved for long periods but is not accessed frequently.
However, storing information in DNA is only half the challenge. Scientists also need to retrieve it quickly and accurately.
Nanopore Sequencing: Fast but Error-Prone
One promising technology for reading DNA is nanopore sequencing. Instead of relying primarily on slower biochemical reactions, nanopore systems detect changes in electrical signals as DNA molecules pass through tiny nanopores.
This can provide rapid and portable DNA readout.
The problem is accuracy.
Nanopore sequencing can produce relatively high error rates, particularly when dealing with shorter DNA fragments. A major source of these errors is insertions and deletions, commonly called indels. These errors can cause the sequence being read to become longer or shorter than the original.
Correcting such errors often requires reading the same DNA information many times. It may also require computationally intensive processes to assemble numerous noisy reads into a reliable sequence.
That creates a problem: if DNA is supposed to become a practical storage medium, retrieving a file should ideally be fast, inexpensive and computationally manageable.
The Problem With Large DNA Storage
Earlier studies attempted to address nanopore errors by creating very large DNA molecules.
For example, researchers have assembled DNA fragments several kilobases long and even hundreds of kilobases long. Such approaches have successfully stored and recovered digital files, including images and videos.
But there is a trade-off.
When sequencing large DNA molecules, the individual reads can be noisy and may begin at different positions. Researchers often need to assemble these reads into longer sequences before decoding the stored information.
Although this is much simpler than biological genome assembly, it can still demand substantial computational resources.
Previous experiments have also required relatively high sequencing coverage—for example, tens or even hundreds of reads covering the same region—to reliably reconstruct stored information.
Chen and his team wanted to reduce both requirements.
A New Strategy: Medium-Length DNA
Instead of using extremely short DNA fragments or very large DNA molecules, the researchers developed a middle ground.
Their approach uses medium-length DNA fragments ranging from several kilobases to tens of kilobases. These molecules can be stored as circular plasmids and later converted into linear DNA fragments for nanopore sequencing.
The researchers tested the system using 28 short circular plasmids of approximately 6–8 kilobases and several longer plasmids measuring roughly 33–43 kilobases.
The DNA encoded information such as Chinese and English poems.
But the most important innovation was not simply the length of the DNA.
It was the coding strategy.
PNC-LDPC: Giving DNA a Built-In Error Correction System
The researchers developed a system called PNC-LDPC, which combines pseudo-noise sequences with low-density parity-check codes.
LDPC codes are powerful error-correction codes already used in modern communication and storage technologies. They add carefully designed redundancy that allows corrupted information to be reconstructed.
The pseudo-noise sequences add another important capability.
They act somewhat like recognizable molecular landmarks. When a nanopore produces a noisy read beginning at an arbitrary location, these sequences can help determine where that read belongs within the original DNA molecule.
More importantly, they can help identify where insertion and deletion errors have occurred.
This means the system does not necessarily need to assemble thousands of overlapping reads into a large contig first.
Instead, noisy reads can be located and corrected more directly.
That is a major computational advantage.
A Single Cleavage Helps Preserve Full-Length DNA
The team also developed an efficient DNA library preparation method.
A key feature is the use of a transposase to make a single cleavage in the encoded DNA. This produces DNA fragments that are close to the full length of the original molecule.
Why does that matter?
If sequencing reads correspond closely to the complete encoded DNA fragment, the system has much more information available from each individual molecule.
The result is a storage system where the researchers can extract useful information without requiring extensive assembly.
Toward Single-Molecule Data Retrieval
The most striking result came from the low sequencing coverage required.
In conventional DNA storage, many copies of the same information may need to be sequenced to overcome errors. Higher coverage means more sequencing, more computational processing and greater cost.
The PNC-LDPC approach substantially reduced this requirement.
Experiments showed reliable recovery at approximately 1.24–3.15× coverage, depending on the data and conditions. With real nanopore sequencing data, the researchers demonstrated recovery at around 3× coverage.
Even more impressively, when the DNA fragments closely matched the designed codewords, error-free recovery was possible at around 1× coverage.
In simple terms, the system can approach a situation where one sequencing observation of a DNA molecule is enough to retrieve the stored information.
That is why the researchers describe the approach as approaching a near single-molecule readout.
Why Low Coverage Matters
Reducing coverage has consequences beyond speed.
Sequencing consumes resources. If a DNA file must be read hundreds of times, the retrieval process becomes expensive and time-consuming.
If the same information can be recovered from only a few reads—or potentially a single molecule—the cost and processing requirements can fall dramatically.
This could be particularly valuable for the write-once, read-many (WORM) model envisioned for DNA archives.
For example, a company could synthesize DNA containing a large archive once and then retrieve different pieces of information many times over its lifetime. In such a scenario, the initial synthesis cost can be shared across many retrieval operations.
As DNA synthesis becomes cheaper, the economics of this model could become increasingly attractive.
What Still Needs to Improve?
The technology is promising, but it is not yet a replacement for conventional hard drives or cloud storage.
One major challenge is the cost of writing DNA. Synthesizing and preparing long DNA molecules remains expensive, especially when compared with producing digital storage on electronic media.
However, DNA synthesis technologies are advancing rapidly. Enzymatic synthesis, for example, aims to produce longer DNA molecules more efficiently and could eventually reduce the cost of creating medium-length storage molecules.
Another challenge is building practical systems capable of storing extremely large archives while maintaining reliable access and physical organization.
A Step Toward Molecular Data Centers
The importance of this research lies in the combination of several ideas: medium-length DNA, powerful error-correcting codes, nanopore sequencing and low-coverage readout.
Rather than relying on massive sequencing depth or computationally expensive assembly, the PNC-LDPC system attempts to make each DNA molecule more informative and easier to recognize.
If DNA synthesis continues to become cheaper and nanopore sequencing becomes faster and more accurate, this strategy could help transform DNA from a biological molecule into a practical molecular storage medium.
The bigger idea is fascinating: future archives may not necessarily sit inside racks of spinning disks or semiconductor memory.
Some of them could exist as carefully designed molecules in tiny containers—waiting to be read when needed.
And with approaches such as PNC-LDPC, researchers are moving closer to a future where retrieving digital information from DNA could be fast, reliable and possible with remarkably little sequencing data.
Reference: Chen, W., Qin, R., Guo, Q. et al. Approaching single-molecule assembly-free readout from medium-length encoded DNA. Nat Commun 16, 10059 (2025). https://doi.org/10.1038/s41467-025-65004-7
Comments
Post a Comment