Krantz, B. | bioRxiv Science

bioRxiv2025

Comparative study of nanopore phenylalanine clamp variants reveals unique peptide biosensing and classification properties

The rapid, label-free, and low-cost detection of peptides is critical for the development of next-generation diagnostics, drug discovery, and environmental monitoring. Nanopore-based biosensing offers a promising platform to address this need by leveraging single-molecule analysis. In this study, we utilize protein engineering to create a series of novel peptide biosensors from the anthrax toxin protective antigen (PA) nanopore by targeting its central phenylalanine clamp constriction (residue F427), a key site known to interact dynamically with translocating molecules. This series of engineered variants were evaluated for their performance in both unsupervised clustering and supervised classification of a diverse set of seven guest-host peptides. Intriguingly, we found that the engineered variants exhibited a broad range of unique biosensing and classification properties. There was a notable divergence between the ability of the variants to intrinsically separate peptides (unsupervised clustering) and their performance in supervised classification tasks. Notably, PA F427A nanopores showed enhanced specificity for small molecular weight peptides that were challenging for WT nanopores to classify, achieving exceptionally high performance (accuracy of 0.93). These findings challenge the assumption that a single unmodified biosensor is sufficient for complex discrimination. Instead, our results highlight the potential for a more robust approach: leveraging the unique, complementary strengths of multiple sensor variants in an ensemble or multiplexed array. Such a system can achieve high and balanced performance across diverse peptide classes, representing a significant step forward in the development of sophisticated nanopore biosensors.

biophysics↗

bioRxiv2025

Deep Learning-Based Classification of Peptide Analytes from Single-Channel Nanopore Translocation Events

Rapid and accurate detection of peptide biomarkers using nanopore biosensors is critical for disease diagnosis and other biomedical applications. Processing large, complex single-channel translocation data streams poses a significant challenge for peptide analyte classification. Here, we present a supervised deep learning data processing pipeline for peptide classification from translocation events. The first stage employs a convolutional and recurrent neural network, adapted from the Deep-Channel multi-channel classifier, to accurately classify raw current recordings into discrete conductance states, including partially blocked sub-conductance intermediates. The second stage, peptide classification, utilizes a novel branched input network with a temporal convolutional network for processing translocation event conductance state sequences and a dense network for incorporating computed event-level and global kinetic features. Using idealized simulated multi-state translocation data for seven peptides, we demonstrate high classification accuracy (0.99) when global features are included alongside event-level features. For classifying mixture samples, where only event-level features are applicable, performance is more modest (0.68 accuracy). Peptide mixture predictions showed reasonable accuracy (MAE 0.045-0.161), although misclassification resulted in false positives. Event stochasticity and the fact that some peptides possessed similar kinetic parameters posed challenging for event-level prediction. However, vote aggregation from translocation event streams achieves perfect 100% accuracy, when predicting pure peptide samples. This proof-of-concept study demonstrates a robust deep learning framework for nanopore peptide classification using simulated data, laying the groundwork for classifying peptides from complex mixtures using real experimental data with the anthrax toxin protective antigen nanopore. HighlightsO_LICreated nanopore biosensor peptide classification pipeline using deep learning. C_LIO_LISequences of discrete conductance state intermediates and features were learned. C_LIO_LIAccurate identification of pure peptide translocation streams via vote aggregation. C_LIO_LIIndividual translocation event classifications can be used to predict peptide mixtures. C_LI

biophysics↗