Study · 2026-10-08
V9: consensus V2 and adaptive computational coverage
Simulated
Five pre-registered hypotheses on decoding under a nanopore-like stress channel: a native consensus kernel, the oracle gap, a better consensus (candidate E), adaptive coverage, and a latent-class channel model. Four accepted, one rejected.
H1 (native kernel, byte-identical and faster): accepted. H2 (oracle gap at coverage 3 and 5): accepted. H3 (a consensus candidate beats V8 with 0 false success): accepted, winner E. H4 (adaptive coverage uses fewer reads with false-termination upper bound ≤ 0.05): accepted. H5 (a latent-structure channel model passes the gating metrics): rejected.
- Classification
- Software-generated DNA strands passed through a software channel model. Not a laboratory result.
- Methodology
- Pre-registration committed before any EVAL decode, with amendments A1 and A2. Exact McNemar with Holm correction; Wilson intervals; paired bootstrap.
- Datasets
- Simulated reads from D13-F1; seeds DEV 90000-90009, EVAL 91000-91099, envelope 92000-92029, scale 93000-93029, benchmark 94000-94009.
- Results
- E: 283 / 600 EXACT vs 240 for V8 (Holm-adjusted p = 5.2e-12). Adaptive coverage: 0 false terminations; 41.7 % / 31.2 % fewer reads. Coverage 3: unsolved.
- Environment
- Shared 4-core / 8-thread Xeon Gold 6240 VPS, CPU only.
- Limitations
- D13-F1 is not a validated channel model. No physical DNA.
- Reproducibility
- experiments/v9/reproduce.sh spot | decoder | coverage | scale
- Source
- https://github.com/vishnuselvam101-hash/VSnexustechnologies-/blob/v9.0.0/docs/V9_COMPLETION_REPORT.md
- References