A recall-rescue fuzzy layer over paired Z-DNA Hunter scans for fast Z-DNA detection.

Z-DNA-Fuzzy Hunter 1.1.0-a detects Z-DNA-forming regions on the audited 385-locus Shin analysis with balanced accuracy 0.883, including recall 0.862 and specificity 0.904.

• Why it matters: Sequence-based Z-DNA detection is constrained by sensitivity–specificity trade-offs, motivating a faster, more interpretable method that can reliably flag Z-DNA-forming regions.
• What they did: The method runs two paired Z-DNA Hunter scans (model 2) with permissive settings (minimum length 6 bp, threshold 30%) and stringent settings (minimum length 10 bp, threshold 60%), then fuses them via a fuzzy expert layer using sequence signal, cross-scan support, genomic context, and bias penalties.
• The result: On the 385-locus Shin dataset, it achieves balanced accuracy 0.883 (recall 0.862, specificity 0.904), comparable to the published Z-DNABERT operating point, and alternative modes reach specificity 0.973.

AI summary of the abstract · may contain errors · not medical advice · read the paper.