Data Sheet 10_Bioinformatic prediction, qCLASH, and AlphaFold3 reveal non-overlapping layers of the miR-214-3p–mRNA interactome.pdf

Nevertheless, all observed overlaps were significantly enriched relative to random expectation (O/E ratios 24.52–80.6; adjusted q < 0.05), and the 95-gene consensus fell far outside the simulated null distribution (P ∼ MC∼ < 10⁻⁶). AlphaFold3 produced high-confidence models (ipTM ≥0.6) for 80.95% of qCLASH-derived interactions, but a substantial fraction of bioinformatically predicted sites failed to form structurally plausible complexes.</p>Discussion<p>Bioinformatic prediction, qCLASH, and AlphaFold3 thus capture distinct and complementary layers of miRNA targeting (sequence-defined potential, context-dependent occupancy, and spatial feasibility) rather than converging on a single landscape.

High AlphaFold3 confidence scores, for instance, did not guarantee biologically functional conformations; even models meeting the thresholds often displayed non-canonical architectures incompatible with silencing. In our mouse lung miR-214-3p model system, our findings support an integrative approach: method selection should align with the specific research question, and multi-layer integration is key to constructing reliable miRNA-target maps in a given biological context.</p>