piaso-tools 1.2.2

PIASO v1.2.2 — tissue-image overlays, trans co-specificity, spatial tutorials

on Python PyPI

Added

• Tissue images under spatial embeddings. pl.embedding(..., basis='spatial', image=True, img_key='hires', image_alpha=, image_crop=)
  and the same on pl.plot_embeddings_split. One resolver serves both
  backends — AnnData uns['spatial'] and a cytome's stored images
  (cytome ≥ 0.2.6; a path to a .cytome works too). The image is drawn in
  full-resolution coordinate units with y increasing downward, so spot
  coordinates overlay with no scaling and no manual axis inversion; plots
  without an image keep the existing y-up behaviour exactly.
  • Library selection is never silent: image=True auto-selects only when
    one library is unambiguous, otherwise it raises naming the options;
    split panels that would mix libraries are drawn without an image, with
    one warning.
• tl.cospecificity_trans is now part of the distribution — trans
  co-specificity between candidate feature pairs (e.g. a cistrome's
  TF→target map) from COSG cell-type specificity profiles; the regulon
  assembly in the cytorete package builds on it.
• Three spatial tutorials on piaso.org: Xenium human breast cancer with
  morphology overlay and ROI queries (run on the public 10x dataset),
  MERFISH multi-section, and Stereo-seq embryo.

Notes

• Feature-detection throughout: older cytome files (or cytome < 0.2.6)
  degrade to image-less plots; no pin changes required.

Suite: 918 passed, 170 skipped.