FusionSieve public validation kit
================================

Version: 1.1.0.dev0 research correction build
Runtime source SHA-256:
8d9ec7e47445917fda75422e9e0152ead8e0b9c08a34bd3703d21de80b0be35c

Purpose
-------
This kit lets a scientist or software reviewer install the corrected
FusionSieve source and run its self-contained software tests. The tests check
specified program behavior. They do not establish clinical sensitivity,
specificity, predictive value, or limit of detection.

Quick validation
----------------
Windows: double-click VALIDATE_ON_WINDOWS.cmd, or run it from Command Prompt.
macOS/Linux: run "sh validate_on_unix.sh".

Each launcher creates a local .validation-venv directory, installs
source[dev], runs the full test suite, and writes validation-report.txt.
The recorded reference run was 143 passed, 1 expected skip, 8 parameterized
subtests, and 0 failed on 2026-07-30. Your current report is authoritative for
your machine; a different test count or any failure requires investigation.

Contents
--------
source/                  Corrected Python source and self-contained tests
wheels/                  Fresh wheel built from the included source
validation-results.json  Aggregate-only evidence from the scientific review
config_no_pon.relative.json
                         Safe example configuration with automatic reporting
                         disabled because no PoN or validated model is supplied
FILE_HASHES.sha256       Integrity hashes for all other files in this kit

The GRCh37 panel sequence and metadata used for the reviewed rerun are included
under source/reference_hg19. No raw sequencing data, sample identifiers,
sample-level results, panel of normals, trained model, patient report, or legacy
release artifact is included.

Scientific status
-----------------
FusionSieve remains research software. The reviewed real-data cohort had no
independent truth labels. Candidate and REVIEW outputs are hypotheses, not
diagnoses. Confirm any candidate by read-level review and an orthogonal method,
and complete assay-specific validation before any patient use.
