Enhanced Lipidomics via False-Positive Filtering and Multilevel Interpretation Reveals Six Mediators Linking...

False-positive annotations in untargeted lipidomics limit biological interpretation. A modified equivalent carbon number (mECN) rule was developed with independently fitted parameters to overcome the fixed coefficient limitation of the classical ECN model, enabling quantifiable false-positive filtering. Validated with a fish oil of known composition and five independent data sets, this workflow improved precision by 8.4% over traditional methods, with 0.5% recall loss and 3.4% F1 gain. Multilevel verification integrating structural (subclass, lipid species, fatty acid) and analytical (metabolite, pathway, network) dimensions was established to strengthen biological interpretation. Applied to a neuropathic pain model, six lipids (CE 20:5, LPC 20:5, LPE 20:5, PG 22:6_22:6, PE 16:1_22:6, and LPI 22:6) were associated with pain modulation, implicating fatty acid elongation/desaturation and phospholipase A2-mediated acyl chain remodeling. This study provides a robust analytical framework for untargeted lipidomics and mechanistic insight into ω-3 nutritional intervention for pain management.