Fixed local competition enrichment neighbor counts. closepairs() returns
both orientations of a pair, and PANORAMIC now retains each pair once before
adding it to the neighborhoods of its two endpoints.
Added stat = "local_comp_global_enrichment" as an explicit alternative
local-composition statistic. It uses a sample-wide target proportion
(a random-label null conditional on observed locations), while the corrected
edge-corrected CSR statistic remains the default.
Fixed block-bootstrap handling: empty tiles no longer contribute artificial
zero-valued block means, block-weighted estimates are invariant to point
input order, and boot = "block" defaults to a 4-by-4 tile grid when tile
counts are not supplied.
Hardened input validation for sample identifiers, preparation parameters, and degenerate convex-hull windows.
Build-system compatibility:
knitr::knitr engine and added
markdown to Suggests so R CMD build does not require Pandoc on
Bioconductor builders.Documentation updates:
DESCRIPTION, README.md, inst/CITATION, and the tutorial
vignette to reflect the published Bioinformatics article
(10.1093/bioinformatics/btag546) and current Bioconductor
installation guidance.panoramic_analyze() as the canonical end-to-end workflow.panoramic() now delegates to panoramic_analyze() and returns a unified
list with prep, stats, pooled, and tables.meta_model, tau2, group_method, keep_boot).panoramic_meta() and panoramic_compare_groups().panoramic_meta_mv() as the primary pooling API, including
single-stage two-group contrasts (beta_diff, se_diff, z_diff,
p_diff, fdr_diff) and optional group-specific heterogeneity estimates.panoramic_extract_contrast() for extracting spatialstats, meta, or
contrast tables.metadata(se)$panoramic$tables.panoramic_sample_deviation(), panoramic_interpret_meta_mv()) from the
public API to reduce surface area.panoramic_spatialstats() internals for robustness (single-sample
support, clearer per-sample/pair failure handling, and optimized local
composition enrichment caching/bootstrapping).plot_representative_samples() and updated plot_volcano(),
plot_forest(), and network plotting helpers to align with multilevel
contrast outputs.Depends: R (>= 4.6),
CITATION parsing fixes, .gitignore final newline).panoramic_prepare() now performs explicit schema validation
(design$sample/design$group, duplicate sample IDs) and stronger
coordinate/window checks after filtering.min_presence.panoramic_meta_mv() now stops on ambiguous group labels after
sanitization (make.names) to avoid silent group collisions.panoramic_spatialstats() now degrades to per-sample NA outputs when local
composition precompute fails, instead of aborting the full run.create_spatial_network() now validates required rowData columns and
documents/guards nonstandard sig_operator = "gt" behavior.plot_forest(), plot_representative_samples()), and network man pages
are synced to current function signatures.