Malathion Aptamers-based Pattern Recognition for Matrix Effect Discrimination in Fruits and Vegetables

Malathion is a commonly used organophosphorus pesticide with low acute toxicity, but overuse leads to residue accumulation and potential health risks. In this study, we introduced an integrated strategy combining magnetic nanoparticle-assisted capture-SELEX with a pattern recognition-based colorimetric aptasensor array for matrix-aware identification of malathion in fruits and vegetables. Four malathion-specific aptamers were first obtained via capture-SELEX, exhibiting differentiated apparent binding affinities ranging from 0.61 to 0.75 μg/mL. Rather than functioning as quantitative probes, these aptamers were deliberately implemented as an ensemble of identification elements to generate distinct gold nanoparticles (AuNPs) colorimetric fingerprints. These response variations arise not only from aptamer-target binding, but also from matrix interference with aptamer-AuNP interactions and potential cross-reactivity between the matrix and both aptamers and AuNPs. By supervised linear discriminant analysis (LDA), eighteen fruit and vegetable matrices spiked with malathion (0.3, 0.5, 0.7 μg/mL) enabled complete classification across all response profiles (four aptamers × eighteen matrices × five replicates). Additionally, 36 unknown samples were consistently and accurately assigned to their corresponding matrix categories, demonstrating the capability to detect malathion while accounting for matrix effects. This pattern aptasensor represents an alternative to conventional lock-and-key biosensors, providing a robust strategy for reliable malathion identification in complex food matrices. A “matrix-target” holistic recognition-based pattern aptasensor for malathion identification integrated with matrix effect discrimination in fruits and vegetables. • An aptamer-ensemble pattern recognition strategy for malathion identification. • Matrix-aware colorimetric fingerprints enabled holistic “matrix-target” recognition. • Complete classification of 90 spiked samples and 36 unknowns by LDA.