NEWS

thanatome 0.99.0

First version submitted to Bioconductor.
thanatome_score(): per-sample signature score, scaled position, state label and technical flags along the ventilator and agony axes, plus the ischemia component of the ventilator axis, with a GTEx tissue (recount3 quantification) or a cohort as reference; method = "regression" adds the joint coefficients and the fraction of deviation explained.
thanatome_score() leaves out the genes a cohort does not express (drop_unexpressed, cohorts of at least 5 samples) and, when x is the unfiltered transcriptome (at least 10 000 and 95% of the genes expressed in the reference), quantile-normalises it to the reference by default (normalize = "auto", with a message; "none" turns it off); thanatome_references carries the statistics of the quantile-normalised GTEx samples (qn).
thanatome_score() and contamination_flags() take tissue, which switches off the flag that is physiological in that tissue (surfactant in lung, pancreatic enzymes in pancreas), also with a cohort reference. Cohort references check their poles (min_pole, default 8 samples per pole, and an internal AUC >= 0.6); unusable poles give NA positions with a warning and are recorded in metadata()$poles.
compare_controls() takes design (covariates of the user's DE, the group term added last), BPPARAM (BiocParallel; random subsets are drawn before fitting, so results do not depend on the backend) and a DGEList.
select_samples(): preset = "no_agony" excludes only samples labelled postmortem_agony and warns when the agony axis is unavailable; the agony interval applies only to non-perfused samples; intervals are validated.
annotate_de() maps HGNC symbols with org.Hs.eg.db and reports the matches; annotate_mra() defines the sign of direction_<axis> per axis and takes padj_threshold.
thanatome_score() accepts a limma EList, ignores the normalisation factors of a DGEList (as the references do), rejects duplicated sample names, and drop_unexpressed now leaves out only genes absent from the cohort (zero counts in most samples).
add_thanatome_score() returns the SummarizedExperiment with the score in its colData and the score metadata in metadata(x)$thanatome.
death_gene_sets() exports the signatures as signed gene sets (named list or GSEABase GeneSetCollection) for gene-set tools.
select_samples(), plot_thanatome() and death_covariates() for sample selection, display and adjustment; contamination_flags() and hardy_layers() as helpers.
annotate_de() and as_de_table() for annotating an existing differential-expression table.
Data: thanatome_signatures (consensus signatures), thanatome_references (per-tissue references and poles, from recount3) and thanatome_example (74 GTEx ovary controls and 40 TCGA-OV tumours from recount3); accessors death_signatures() and death_reference(); build scripts in inst/scripts/.
compare_controls() repeats the DE with explicit control subsets against random subsets of the same size; annotate_mra() annotates regulator tables; audit_report() writes the summary. The DE inside compare_controls() follows the edgeR 4 quasi-likelihood pipeline (glmQLFit() without estimateDisp()).