菜单
Cell-cell communication analysis for spatial transcriptomics using LIANA, CellPhoneDB, FastCCC, or CellChat, with method-specific parameter hints and standardized ligand-receptor outputs.
默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。
决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/mdbabumiamssm/LLMs-Universal-Life-Science-and-Clinical-Skills- --skill spatial-communication命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
基于 SOC 职业分类
正在显示 SKILL.md
| name | spatial-communication |
| description | Cell-cell communication analysis for spatial transcriptomics using LIANA, CellPhoneDB, FastCCC, or CellChat, with method-specific parameter hints and standardized ligand-receptor outputs. |
| version | 0.5.0 |
| author | OmicsClaw Team |
| license | MIT |
| tags | ["spatial","communication","ligand-receptor","liana","cellphonedb","fastccc","cellchat"] |
| metadata | {"omicsclaw":{"domain":"spatial","allowed_extra_flags":["--cell-type-key","--method","--species","--liana-expr-prop","--liana-min-cells","--liana-n-perms","--liana-resource","--cellphonedb-iterations","--cellphonedb-threshold","--fastccc-single-unit-summary","--fastccc-complex-aggregation","--fastccc-lr-combination","--fastccc-min-percentile","--cellchat-prob-type","--cellchat-min-cells"],"param_hints":{"liana":"[Truncated]","cellphonedb":"[Truncated]","fastccc":"[Truncated]","cellchat_r":"[Truncated]"},"legacy_aliases":["communication"],"saves_h5ad":true,"requires_preprocessed":true,"requires":{"bins":"[Truncated]","env":"[Truncated]","config":"[Truncated]"},"emoji":"📡","homepage":"https://github.com/TianGzlab/OmicsClaw","os":["macos","linux"],"install":["[Truncated]"],"trigger_keywords":["cell communication","cell-cell communication","ligand receptor","ligand-receptor","LIANA","CellPhoneDB","FastCCC","CellChat"]}} |
You are Spatial Communication, the OmicsClaw skill for ligand-receptor interaction analysis in spatial transcriptomics data. The skill exposes four backends with different statistical assumptions and now keeps their core parameters method-specific instead of flattening them into a generic interface.
liana.mt.rank_aggregate.iterations and threshold controls.statistical_analysis_method API.skills/spatial/_lib/viz primitives.ligand, receptor, source, target, score, pvalue.reproducibility/commands.sh.| Format | Extension | Required Fields | Example |
|---|---|---|---|
| AnnData (preprocessed) | .h5ad | X (log-normalized), obsm["spatial"], obs["leiden"] or another cell type column | processed.h5ad |
Current OmicsClaw spatial-communication uses log-normalized expression in
adata.X for all four backends. Do not pass z-scored or centered matrices.
| Method | Input Matrix | Notes |
|---|---|---|
liana | adata.X; uses adata.raw if available | adata.raw is treated as the log-normalized full gene space, not raw UMI counts |
cellphonedb | adata.X | CellPhoneDB docs explicitly warn against transforms that convert zeros into non-zero values |
fastccc | adata.X | Current OmicsClaw wrapper writes the AnnData view to h5ad and runs FastCCC on that matrix |
cellchat_r | adata.X | CellChat tutorial expects normalized, log-transformed expression |
Current wrapper support is intentionally method-specific:
| Method | Supported species in current OmicsClaw wrapper |
|---|---|
liana | human, mouse |
cellphonedb | human only |
fastccc | human only |
cellchat_r | human, mouse |
The CLI only exposes human and mouse. Unsupported combinations fail fast.
adata.uns["ccc_results"] and a method-specific result key.plot_communication() and plot_features() building blocks.figure_data/ CSVs and manifests for optional downstream R-side customization.report.md, result.json, processed.h5ad, tables, figure_data/, and reproducibility metadata.# Default OmicsClaw CLI alias
oc run spatial-cell-communication \
--input <processed.h5ad> --output <report_dir>
# LIANA with method-specific controls
oc run spatial-cell-communication \
--input <processed.h5ad> --method liana \
--cell-type-key cell_type --species mouse \
--liana-resource auto --liana-expr-prop 0.1 --liana-min-cells 10 --liana-n-perms 1000 \
--output <dir>
# CellPhoneDB statistical method
oc run spatial-cell-communication \
--input <processed.h5ad> --method cellphonedb \
--cellphonedb-threshold 0.1 --cellphonedb-iterations 1000 \
--output <dir>
# FastCCC
oc run spatial-cell-communication \
--input <processed.h5ad> --method fastccc \
--fastccc-single-unit-summary Mean \
--fastccc-complex-aggregation Minimum \
--fastccc-lr-combination Arithmetic \
--fastccc-min-percentile 0.1 \
--output <dir>
# CellChat via R
oc run spatial-cell-communication \
--input <processed.h5ad> --method cellchat_r \
--species mouse --cellchat-prob-type triMean --cellchat-min-cells 10 \
--output <dir>
# Demo mode
oc run spatial-cell-communication --demo --output /tmp/comm_demo
# Direct script entrypoint
python skills/spatial/spatial-communication/spatial_communication.py \
--input <processed.h5ad> --method liana --output <dir>
Every successful standard OmicsClaw wrapper run, including oc run and
conversational skill execution, also writes a top-level README.md and
reproducibility/analysis_notebook.ipynb to make the output directory easier
to inspect and rerun. Direct script execution primarily produces the
skill-native outputs plus reproducibility/commands.sh.
liana.mt.rank_aggregateresource_name, expr_prop, min_cells, n_permsauto -> consensus for human, auto -> mouseconsensus for mousecpdb_statistical_analysis_method.calliterations, thresholdfastccc.statistical_analysis_methodsingle_unit_summary, complex_aggregation, LR_combination, min_percentilecomputeCommunProb + filterCommunication through sc_cellchat.Rtype, min.cellsCellChat, SingleCellExperiment, and zellkonverteroutput_directory/
├── README.md # wrapper mode
├── report.md
├── result.json
├── processed.h5ad
├── figures/
│ ├── lr_dotplot.png
│ ├── lr_heatmap.png
│ ├── lr_spatial.png # if spatial score layer exists
│ ├── communication_roles_spatial.png
│ ├── communication_hub_umap.png # or communication_hub_spatial.png
│ ├── signaling_roles.png
│ ├── source_target_summary.png
│ ├── communication_pvalue_distribution.png
│ ├── communication_score_vs_significance.png
│ └── manifest.json
├── figure_data/
│ ├── lr_interactions.csv
│ ├── top_interactions.csv
│ ├── communication_summary.csv
│ ├── signaling_roles.csv
│ ├── source_target_summary.csv
│ ├── communication_run_summary.csv
│ ├── communication_spatial_points.csv
│ ├── communication_umap_points.csv
│ └── manifest.json
├── tables/
│ ├── lr_interactions.csv
│ ├── top_interactions.csv
│ ├── communication_summary.csv
│ ├── signaling_roles.csv
│ ├── source_target_summary.csv
│ ├── cellchat_pathways.csv # CellChat only
│ ├── cellchat_centrality.csv # CellChat only
│ ├── cellchat_count_matrix.csv # CellChat only
│ └── cellchat_weight_matrix.csv # CellChat only
└── reproducibility/
├── analysis_notebook.ipynb # wrapper mode
├── commands.sh
├── requirements.txt
└── r_visualization.sh
figures/manifest.json describes the
standard OmicsClaw communication story for routine analysis delivery.figure_data/ is the bridge layer: downstream plotting code should read
exported CSVs instead of rerunning LIANA, CellPhoneDB, FastCCC, or CellChat.skills/spatial/spatial-communication/r_visualization/ contains starter
templates that consume figure_data/ and write polished figures under
figures/custom/.Required (Python):
scanpyOptional (Python):
lianacellphonedbfastcccOptional (R):
CellChatSingleCellExperimentzellkonverterTrigger conditions:
Chaining partners:
spatial-preprocess: prepares log-normalized AnnData inputspatial-annotate: improves cell type labels before communication analysisspatial-domains: helps interpret communication within spatial regions