2026-10-23 SNP inclusion diagnostic; post-v0.9, exploratory.
Hold 256 arrays, ENmix package, oob/RELIC/RCP and quantile1 fixed.
Native rgDataSet conversion uses the installed manifest, raw measured channels
and actual bead counts. Validate exact common raw M/U values and native
SNP-excluded corrected intensities against the saved v0.9 MethylSet fixture.
The included branch changes 25 type-I and 40 type-II SNP loci. This affects
out-of-band estimation and intensity fitting as well as quantile targets.
Primary diagnostic: beta and score discrepancy against deposited matrix,
not favorable repeatability. No outcome-dependent array/probe exclusions.
Validate excluded/included raw controls and shared assay addresses.
If fixture equality fails, identify the discrepancy before whole-cohort run.
If it passes, reuse v0.9 excluded caches, process included branch and normalize
all 256 arrays together. Check required measured inputs and all clock scores.
Residual errors prevent claiming exact reconstruction; isolate interpretation.
