Amplification-free detection of mycoplasma pneumoniae via CRISPR-Cas12a and deep learning-optimized crRNAs on a...

Accurate and rapid diagnosis of Mycoplasma pneumoniae infection is essential for reducing its significant health burden. An amplification-free CRISPR-Cas12a-mediated detection platform has been developed, incorporating a deep learning-optimized crRNA library (CCDLCL) targeting conserved regions of the MP P1 gene. The system enables visual readout via lateral flow strips, supporting its potential as a point-of-care testing (POCT) nucleic acid testing strategy. Through computational design and screening, 16 highly active crRNAs were identified from an initial set of over 50 candidates. Combinatorial use of these crRNAs demonstrated synergistic enhancement of fluorescence signal intensity and reaction kinetics. Compared to single-crRNA assays, the multiplexed crRNA library improved sensitivity by 16.8-fold, achieving a limit of detection (LOD) of 0.15 pM, and reduced time to signal saturation by 30 %. When deployed on lateral flow strips, the assay exhibited a tenfold increase in visual detection sensitivity, with a LOD of 100 pM. Clinical evaluations confirmed high specificity-showing no cross-reactivity with SARS-CoV-2, hepatitis B virus (HBV), or human genomic DNA-and over 95 % agreement with standard clinical results without target pre-amplification, delivering outcomes within 45 min. This study establishes a deep learning-facilitated crRNA design framework and a novel crRNA library-based detection system, offering a feasible approach for POCT nucleic acid testing in resource-limited settings and paving the way for streamlined clinical translation of CRISPR-Cas diagnostics.