Use this skill when the user is analyzing 10x Genomics Xenium in situ data, or files such as transcripts.csv, transcripts.parquet, cells.csv, cell_feature_matrix.h5, or Xenium output directories. Use it for data preparation, preprocessing, differential expression, cell type annotation, spatial analysis, domain detection, and cell segmentation.
Use this skill for Xenium platform detection, workflow order, step execution, and helper-library usage.
Xenium or 10x Xeniumtranscripts.csv, transcripts.parquet, cells.csv, or cell_feature_matrix.h5Xeniumanalysis.tar.gz from Xenium preprocessingIf the platform is still unclear, ask before executing platform-specific steps.
.h5ad already present in the attached data?analysis.tar.gzIf the environment supports redeem_package, install the Xenium package before analysis:
package_code: 7a4f4bd980b3739a825072a975dd9a376c267ff7c84c1c9c59c8da196e58c3bdpackage_version_id: 401If redeem_package is unavailable, continue with the repo-local docs and the current environment.
Read main.md for the step plan, then load each step doc before executing it.
.h5ad and viewer-ready assets — step details.h5ad, attach spatial tiles, and render with w_h5 — step detailsIf a step requires Xenium helper code, import from the skill's lib/ directory:
import sys
sys.path.insert(0, "<skill-root>/lib")
from xenium_cell_type import ...
Resolve <skill-root> to the directory where this skill is checked out in the current environment.
w_plot. Do NOT reuse the variable name fig — use descriptive names like fig_qc, fig_umap.If latch-workflows, latch-plots-ui, or latch-data-access are available, prefer them for:
If those sibling skills are not available, use the local wf/, steps/, and README.md docs directly.