Package: tewas 0.99.2

tewas: Unit-Level Multiple Testing for Epigenome-Wide Association Studies with an Anchor-Calibrated Knowledge Prior

A multiple-testing engine for DNA methylation arrays (engine v1.6) and a frame layer for the designs of behavioral epigenetics. Probes are grouped into genomic-run units; each unit is tested with four complementary statistics (dense sum, coherence-gated flip, Simes, kernel matched filter) whose null distributions come from one shared permutation stream with a calibrated exponential tail below the permutation floor; the arm p-values are combined by the Cauchy combination and units are selected by weighted Benjamini-Hochberg with cross-fitted structural weights and, optionally, a per-probe knowledge prior from longitudinal within-person change measured in an external cohort. Anchor probes carried in the scan measure whether the prior is licensed (the lens) and whether the scan's null is calibrated (the inflation guard, read on an alive null ruler); the prior is tempered toward zero when it is not, and nothing is corrected. The frame layer declares a design (moderation, curvature, threshold, acceleration, methylation as predictor or as outcome), runs every arm through the same engine, merges selected units into regions with coordinates, annotates the probes with their target genes, and writes tables, trajectory figures and a self-declaring summary. Unit-level FDR is controlled under the stated assumptions; probe-level false discovery and the calibration of every arm were measured in registered simulations whose readings the package prints beside its results. The anchor panel and the alive ruler are shipped under inst/extdata as per-probe summaries only. Input is a beta matrix or a SummarizedExperiment (a minfi GenomicRatioSet included); selected regions, units and probes are returned as GRanges on request.

Authors:Tsachi Ein-Dor [aut, cre]