Abstract: Xenobiotic nucleic acids (XNA) are nucleic acid analogues not present in nature that can be used for the storage of genetic information. In vivo XNA applications could be developed into novel biocontainment strategies, but are currently limited by the challenge of developing XNA processing enzymes such as polymerases, ligases and nucleases. Here, we present a structure-guided modelling-based strategy for the rational design of those enzymes essential for the development of XNA molecular biology. Docking of protein domains to unbound double-stranded nucleic acids is used to generate a first approximation of the extensive interaction of nucleic acid processing enzymes with their substrate. Molecular dynamics is used to optimise that prediction allowing, for the first time, the accurate prediction of how proteins that form toroidal complexes with nucleic acids interact with their substrate. Using the Chlorella virus DNA ligase as a proof of principle, we recapitulate the ligase's substrate specificity and successfully predict how to convert it into an XNA-templated XNA ligase.