Research HighlightPublished: 04 September 2026Molecular engineeringArunima Singh1 Nature Methods volume 23, page 1679 (2026) Cite this articleSave articleView saved researchThe team first mapped the genome-wide off-target landscape of adenine base editors (ABEs) using a new sequencing method, deoxyinosine (dI)-profiling, which traces the deoxyinosine intermediate created during editing. They then fed both on-target and off-target DNA sequences into AlphaFold3, discovering that the model’s internal ‘contact probability’ output, rather than its predicted 3D structures, was far more sensitive at revealing subtle differences between correctly and incorrectly bound protein–DNA–RNA complexes. By correlating these contact probabilities with real off-target editing signals, the authors identified clusters of Cas9 residues, termed consensus contact regions, whose contact behavior consistently shifted at off-target sites. Mutating specificity-determining residues within these regions yielded editors with substantially reduced off-target activity while preserving on-target efficiency. When the authors applied this strategy to Cas9 and its TadA8e deaminase, the best resulting variant outperformed several existing high-fidelity ABEs across genome-wide profiling, targeted sequencing, RNA off-target assays and R-loop measurements. The framework generalized beyond ABEs and was shown to also improve a Cas12a-based cytosine base editor.“For basic research, we would imagine ContactSeek could help us better understand how these DNA-binding proteins and RNA-binding proteins recognize their targets. For therapeutics, ContactSeek could help optimize oligonucleotide therapeutics, such as ASOs [antisense oligonucleotides] and siRNAs [small inhibitory RNAs], and antibody drugs,” highlights Yi. The team hopes to improve the framework to design combinations of mutations and understand how they work synergistically in the future.This is a preview of subscription content, access via your institutionAccess options Access through your institutionAccess Nature and 54 other Nature Portfolio journalsGet Nature+, our best-value online-access subscription27,99 € / 30 dayscancel any timeLearn moreSubscribe to this journalReceive 12 print issues and online access269,00 € per yearonly 22,42 € per issueLearn moreBuy this articlePurchase on SpringerLinkInstant access to the full article PDF.39,95 €Prices may be subject to local taxes which are calculated during checkoutSubjectsBiotechnologyComputational modelsMolecular engineeringProtein designAuthor informationAuthors and AffiliationsNature Methods https://www.nature.com/nmeth/Arunima SinghAuthorsArunima SinghView author publicationsSearch author on:PubMed Google ScholarCorresponding authorCorrespondence to Arunima Singh.Rights and permissionsReprints and permissionsAbout this article