We leveraged our library by employing our data-augmentation technique based on splicing XNA/DNA signals to generate reads containing UBs in all contexts. Using these reads we trained a generalizable model capable of sequencing UBs with high accuracy (>80%) and specificity (99%).

To obtain a ML model that can deconvolve these signals and basecall UBs along with natural bases, we developed an enzymatic framework to synthesize with high XNA purity (>90%) a library of 1,024 UB-containing oligonucleotides representing all 6,144 single-UB 6-mer contexts.

We validated that the Nanopore sequencer is capable of processing XNAs containing the Px-Ds unnatural base pair and assessed the distinguishability of their measured signals versus DNA control and the canonical bases.

The inability to read XNAs in a direct manner was a significant limitation for xenobiology. Here we demonstrate that XNA-containing templates can be directly sensed using a MinION sequencer to generate unique signals that can be deconvoluted by a specialized basecaller model.