Digital Decoding of Multicomponent Protein Systems via Nanocavity-Confined Single-Molecule Raman Fingerprinting

The ability to identify individual protein species within multicomponent systems remains a major challenge, yet is essential for next-generation molecular diagnostics and proteomic analysis. Here, we present a single-molecule Raman fingerprinting strategy for automatic digital decoding of protein compositions in complex systems. A dual-amplified, interface-coupled plasmonic nanocavity architecture synergistically integrates gap-mode coupling with surface plasmon resonance, generating ultralow-volume, highly enhanced hotspots that reproducibly confine and isolate single proteins, enabling acquisition of intrinsic Raman spectra free from spectral overlap. Using this platform, we achieve high-throughput hyperspectral Raman fingerprinting of seven representative proteins. A customized machine learning algorithm trained on single-molecule Raman datasets enables automatic identification and spatial mapping of individual protein species, yielding quantitative and addressable decoding maps. This broadly applicable strategy establishes an intelligent, data-driven framework for multiplexed protein analysis under ambient conditions, with far-reaching implications for molecular diagnostics, biosensing, and mechanistic studies of protein function.