Nature Biotechnology editors compared four methods for writing DNA and identified the production of long, accurate sequences as a bottleneck

Nature Biotechnology editors compared four methods for writing DNA and identified the production of long, accurate sequences as a bottleneck

On August 19, an editorial review in Nature Biotechnology compared four approaches to writing DNA. It explains why producing long, accurate sequences slows the transition from computational design to laboratory testing of a biological idea.

Computational models can propose thousands of protein and genetic construct designs within hours. Before a laboratory can test any of them, it must obtain the corresponding DNA sequence and confirm its accuracy. This physical stage makes the progression from design to experiment slow and expensive.

The industry standard is phosphoramidite chemistry. It adds nucleotides, the chemical letters of DNA, one at a time to produce short fragments called oligonucleotides, typically up to about 200 nucleotides long. These fragments are then assembled into a longer sequence. Errors accumulate over hundreds of chemical cycles, and assembly cannot correct an error already present in a source fragment.

Sidewinder and MOSAIC operate specifically at the assembly stage. Sidewinder uses a third DNA strand to specify the order in which fragments are joined. This strand remains outside the finished molecule, and the method facilitates the assembly of repetitive sequences. In MOSAIC, overlapping oligonucleotides self-assemble in a single reaction. A cellular DNA repair system then seals the breaks and converts the intermediate into a plasmid, a circular DNA molecule used in genetic engineering.

The assembled construct must still be amplified in a cell. The rapid growth of the bacterium Vibrio natriegens shortens this cloning stage. The review addresses an earlier constraint: before amplification, long DNA must first be written and assembled with acceptable accuracy.

Another approach changes the writing process itself. TdT is an enzyme that adds specified nucleotides one at a time without a DNA template. The review cites a result from Ansa Biotechnologies involving molecules longer than 1 000 bases. Retrons are bacterial genetic elements that carry an RNA template and a reverse transcriptase, which copies RNA into DNA. In the experiments described, retrons produced sequences tens of base pairs long.

The four approaches address different constraints. Conventional chemistry produces fragments. Sidewinder and MOSAIC join them. TdT writes the sequence directly. Retrons use the cell itself. The pace of the experimental cycle depends on achieving sufficient length, accuracy, and speed across all these stages.

For proteins designed with AI, the time required to reach an experiment includes both DNA production and verification. Gene and cell therapies require long, accurate sequences for constructs containing multiple genetic edits and large DNA insertions. The new methods will need to be evaluated using sequences of different lengths and complexity so that their speed, accuracy, and reliability can be compared.