Digital Nanopore Counting Enabling Direct Identification of Trace-Level Neoantigens in Native Biological Samples

Abstract Tumor-specific neoantigens have been the most promising targets for TCR-based cancer therapies; however, their detection is greatly hindered by low abundance, sequence variability, and potential modifications. To address these, we developed a digital nanopore framework to unambiguously identify and effectively quantify candidate neoantigens directly in cell samples via event-resolved molecular identification and counting. By engineering a multirecognition-region nanopore and developing a universal water-osmosis driving force, we simultaneously enhanced the detection throughput and specificity. Combined with enriched current signatures and a gradient-boosting decision-tree classifier, the platform identifies individual ovalbumin-derived model neoantigens with single-amino-acid resolution from digitally assigned events in both exogenously incubated and endogenously expressed cell samples. We show that the limit of quantification is improved to the fmol level in native biological environments, enabling the direct identification of neoantigens from as few as 5000 cells within 30 min. We further demonstrate the identification of a native Epstein–Barr virus-associated neoantigen, highlighting the potential of this approach in ultrasensitive analysis of rare biomolecular species directly in biological samples.