Quantifying XNA replication fidelity using nanopore sequencing

Expanded genetic alphabets built from unnatural base-pairing xenonucleic acids, or ubp XNAs, are part of an emerging frontier in biotechnology, therapeutics, and synthetic biology. However, their adoption has been constrained by replication fidelities that remain substantially lower than those of standard, natural DNA. We present a method for rapidly and accurately measuring ubp XNA replication fidelity using single-molecule nanopore sequencing. By coupling nanopore sequencing with post hoc machine-learning classifiers, we infer strand-resolved, cycle-dependent fidelity parameters directly from PCR amplification data and place these measurements on a rigorous mathematical footing that accounts for coupled replication dynamics. We apply this approach to three chemically distinct ubp systems—hydrogen-bonding B≡S and P≡Z pairs, and the hydrophobic Ds:Diol-Px pair. Using this method, we demonstrate: (i) rapid screening of reaction conditions that yields >98.6% per-cycle replication fidelity for the B≡S base pair, (ii) single-molecule tracking of replication outcomes in an ‘Hachimoji’ 8-letter system (ATGCBSPZ), and (iii) first estimates of context-dependent inversion error rates in the hydrophobic Ds≡Px system, revealing inversion frequencies of ≈8.5% in the 5`-pur-Ds-pur-3` context. Together, this approach provides a general route to quantitatively assess polymerase fidelity in expanded genetic alphabets, enabling systematic optimization across chemistries, enzymes, and reaction conditions.