Transcript provenance and normal-parent sequence comparisons for EMC fusion-junction antisense designs

Versioned computational inputs, attributable sequence references, junction provenance, 40 design rows (35 source-linked and five hypothetical controls), 77 normal-parent transcript records, exact match locations, sensitivity results and reproducible analysis. These are sequence comparisons, not evidence of cleavage, potency, safety or clinical benefit. Downloaded literature is linked through the source ledger, not republished in this archive. This deposition is the archive cited by the manuscript's availability statement. It carries 74 files taken from https://github.com/trimcrae/Rare-cancers at revision 9efe8847afc406370c817afcfa19cc2c8f40c04c, together with the manifest that names and hashes every one of them. Verifying this archive. Every file's SHA-256 is listed in zenodo-manifest.json. That file cannot carry its own hash, so it is recorded here instead: 19cf8ab2ef6bf84d8f1871704b1e720e03a69ee12327b6750f4a07ff4d1e7d4f. The archive's content digest, derived over the file list, is 0d8363a76c6c0d9a05b9235c0cfb21c4c636d0fa085bb2a5e309922ed7f489b9. Reproducing the results offline. Extract the ZIP while preserving its repository-relative directories. Enter research/release-candidates/PUB-ASO/2026-09-26/evidence and run python analyze.py using Python 3.11 or later. No external packages or network are required. Compare SHA-256 hashes for the eleven result files with this manifest; all should reproduce exactly. Read repository-deposit/README.md and supplementary-methods.md for source classes, limitations, licensing and the optional separate read-prefix replay. Research use only. The oligonucleotide sequences in these artefacts are research reagents; nothing here is for administration to any person or animal, and nothing in it asserts efficacy, safety, delivery or clinical readiness.