The central dogma of molecular biology teaches that genetic information travels in one direction: from DNA to RNA to protein. This elegant flow, first articulated by Francis Crick in 1958, has long been held as an inviolable principle of life. Yet a team of bioengineers at Stanford University has now accomplished what could only have seemed paradoxical to the architects of molecular biology — they have found a way to read a protein back into the language of DNA.
Published on March 18, 2026 in Nature Biotechnology, the work of Liwei Zheng, H. Tom Soh, and colleagues describes a strategy they call “reverse translation”: a chemical and molecular process that converts the amino acid sequence of individual peptides into barcoded DNA libraries, which may then be read by existing high-throughput sequencing platforms. The implications for proteomics — and, by extension, for medicine — could scarcely be overstated.
At the molecular heart of the method lies a modified Edman degradation, a classical technique in which the N-terminal amino acid of a peptide is chemically tagged and cleaved, one residue at a time. In the Stanford approach, each amino acid is conjugated, before cleavage, to an azide-modified phenyl isothiocyanate. A DBCO-modified biotinylated primer is then click-coupled to this azide group and extended by DNA polymerase, thereby associating the amino acid with a peptide-specific DNA barcode. Following cleavage, the DNA-barcoded amino acid fragments are enriched by streptavidin pulldown and subjected to proximity extension assay (PEA): DNA-barcoded, amino acid-specific antibodies bind the fragments and, through proximity ligation, generate amplifiable DNA reporters that record both the identity of the amino acid and its position within the original peptide. After each iterative cycle, PCR amplification and standard sequencing decode the entire peptide, residue by residue.
What distinguishes this strategy from prior proteomics approaches is its sensitivity. Conventional mass spectrometry may analyse upwards of one billion protein molecules per run and yet detect fewer than one million; the Stanford method, by contrast, may interrogate approximately one thousand times more molecules from the same sample, with single-molecule resolution. Furthermore, because the process converts the peptide sequence into a DNA output, it may draw upon decades of innovation in nucleic acid technology, like PCR amplification and high-throughput sequencing, tools that have no equivalent in the protein world.
From a therapeutic standpoint, this advance could prove transformative. Rare and low-abundance proteins, which current instruments frequently fail to detect but which may drive disease at the cellular level, might at last become visible. The method is already being explored for applications in immunotherapy: the ability to compare the proteomic profiles of single immune cells — distinguishing, for instance, those that respond to CAR-T therapy from those that do not — could open new avenues for personalized cancer treatment. Post-translational modifications, including phosphorylation events that signal disease states, may also be mapped with unprecedented precision. The technology remains in its early stages and will require further optimization before clinical adoption, but the conceptual architecture is sound, the proof of principle compelling, and the commercial interest already secured.
Nature, it seems, reserves its deepest secrets for those prepared to read her in a borrowed alphabet.
Leave a Reply