MutationSieve software-verification evidence
============================================

SOFTWARE CONTROL GATES: PASS (13/13).
ANALYTICAL AND CLINICAL VALIDATION: NOT PERFORMED.
Version: 0.1.0.dev1
Frozen date: 2026-08-01

This archive contains public derivatives of deterministic aligned-BAM
software-regression fixtures and the exact wheel/source distribution.
These are not FASTQ-to-pinned-production-aligner synthetic-read validation
data. They establish no sensitivity, specificity, PPV, NPV, LoB, LoD95,
precision, reproducibility, transportability, or clinical validity.

Twenty-four of 25 positive MutationSieve-simulator alleles required 67 disclosed
forced top-up observations; the remaining allele reached its floor ordinarily.
Three SpikeForge alleles used a separately implemented BAM edit-output path;
the baseline alignment context remained shared, so this was not an independent experiment.
The 28/29 result is constructed-control recovery, not sensitivity. All selected
calls are REVIEW; PASS/HIGH are disabled and absent. External callers, public
truth, physical samples, and patient samples were NOT RUN.

JUnit: 343 direct cases plus 3 subtests; failures=0, errors=0, skipped=3.
Verify PUBLIC-EVIDENCE-SHA256SUMS.txt and read software-verification-manifest.json.
