[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7EFR7X68PHT099BDA3M579Y":3,"skill-same:01M7EFR7X68PHT099BDA3M579Y":62},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":30,"commitSha":31,"skillDirectory":32,"forksCount":14,"repositoryHtmlUrl":33,"repositoryHomepageUrl":33,"ownerHtmlUrl":34,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":35,"install":36,"resources":49},"01M7EFR7X68PHT099BDA3M579Y","bio-spatial-transcriptomics-spatial-domains",null,"Identify spatially coherent tissue domains (regions like cortical layers, tumor vs stroma) in Visium, Visium HD, Xenium, MERFISH, Slide-seq, and Stereo-seq data with Squidpy, BANKSY, BayesSpace, STAGATE, and GraphST. Use when distinguishing a domain (a region with many cell types) from a cell type (one cell's identity) and a niche (local cell-type composition); choosing a domain method by tissue geometry (laminar\u002Fcontinuous vs high-resolution imaging vs non-contiguous); tuning the spatial-weight knob (BANKSY lambda, BayesSpace smoothing, SpaGCN histology weight, GNN graph radius) to avoid over-smoothing into blobs or under-smoothing into salt-and-pepper; choosing the number of domains k as a biological decision with k+-1 sensitivity; and reading the Yuan 2024 benchmark with the DLPFC continuous-laminar caveat.","Mr-Nobody-Anonymous\u002FAll-skills","Mr-Nobody-Anonymous","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F184253119?v=4","01M7AEXFJHKTYQW5R0ZCQQTWFV","anonymous","01M7C00M3GQR4QKP9GZA42GFAK",0,"MIT","https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous\u002FAll-skills\u002Fblob\u002Fmain\u002Fawesome_skills\u002Fbiology\u002Fbio-spatial-transcriptomics-spatial-domains\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: bio-spatial-transcriptomics-spatial-domains\ndescription: Identify spatially coherent tissue domains (regions like cortical layers, tumor vs stroma) in Visium, Visium HD, Xenium, MERFISH, Slide-seq, and Stereo-seq data with Squidpy, BANKSY, BayesSpace, STAGATE, and GraphST. Use when distinguishing a domain (a region with many cell types) from a cell type (one cell's identity) and a niche (local cell-type composition); choosing a domain method by tissue geometry (laminar\u002Fcontinuous vs high-resolution imaging vs non-contiguous); tuning the spatial-weight knob (BANKSY lambda, BayesSpace smoothing, SpaGCN histology weight, GNN graph radius) to avoid over-smoothing into blobs or under-smoothing into salt-and-pepper; choosing the number of domains k as a biological decision with k+-1 sensitivity; and reading the Yuan 2024 benchmark with the DLPFC continuous-laminar caveat.\ntool_type: python\nprimary_tool: squidpy\nsource: \"https:\u002F\u002Fgithub.com\u002FGPTomics\u002FbioSkills\"\nsource_repository: \"GPTomics\u002FbioSkills\"\nsource_path: \"spatial-transcriptomics\u002Fspatial-domains\u002FSKILL.md\"\nlicense: \"MIT\"\nimported_at: \"2026-09-20\"\n---\n\n## Version Compatibility\n\nReference examples tested with: squidpy 1.4+, scanpy 1.10+, anndata 0.10+, scikit-learn 1.4+; BANKSY (banksy_py \u002F R Banksy), BayesSpace 1.12+ (R), STAGATE\u002FGraphST optional\n\nBefore using code patterns, verify installed versions match. If versions differ:\n- Python: `pip show \u003Cpackage>` then `help(module.function)` to check signatures\n- R: `packageVersion('\u003Cpkg>')` then `?function_name` to verify parameters\n\nIf code throws ImportError, AttributeError, or TypeError, introspect the installed\npackage and adapt the example to match the actual API rather than retrying.\n\n# Spatial Domain Detection\n\n**\"Identify tissue domains in my section\"** -> Partition the tissue into spatially contiguous regions of homogeneous expression\u002Fcomposition, using BOTH transcriptional similarity AND spatial proximity, so the output is a region label per spot\u002Fcell that is spatially coherent.\n- Python: `squidpy.gr.spatial_neighbors` for the graph, then a neighbor-augmented or graph-based domain method (BANKSY, STAGATE, GraphST)\n- R: BayesSpace (`spatialCluster`), BASS, or Banksy\n\n## Governing Principle\n\nA spatial domain is a REGION, not a cell type, and not a niche -- conflating the three is the central conceptual error of this analysis. A CELL TYPE is one cell's transcriptional identity. A NICHE (cellular neighborhood) is the local cell-type COMPOSITION around a cell -- which types co-occur. A SPATIAL DOMAIN is a contiguous tissue region (a cortical layer, tumor core, stromal band) that contains MANY cell types and several niches. If the question is \"which cell types co-occur,\" that is a niche question and belongs to neighborhood-enrichment analysis (spatial-statistics), NOT domain segmentation; running the wrong one answers the wrong question.\n\nDomain methods exist because plain Leiden\u002FLouvain on expression alone ignores coordinates and produces salt-and-pepper, spatially-incoherent labels. Every domain method deliberately adds a spatial term so neighbors tend to share a label. The load-bearing decision is therefore NOT the clustering algorithm -- it is the SPATIAL-WEIGHT knob (BANKSY lambda, BayesSpace smoothing, SpaGCN histology weight, GNN graph radius). Too much spatial weight is the #1 trap, over-smoothing: it erases real boundaries and merges biologically distinct regions into blobs. Too little reverts to salt-and-pepper. The number of domains k is the second decision, and it is a BIOLOGICAL choice (how fine a regionalization the question needs), not a statistic to optimize against a silhouette score -- report k, justify it, and show sensitivity to k+-1.\n\n## Domain vs Niche vs Cell Type\n\n| Concept | What it is | Unit | Right tool |\n|---------|-----------|------|-----------|\n| Cell type | One cell's transcriptional identity (lineage\u002Fstate) | a cell | clustering + markers (single-cell\u002Fclustering) |\n| Niche \u002F cellular neighborhood | Local cell-type composition -- which types co-occur around a cell | a neighborhood | neighborhood enrichment \u002F co-occurrence (spatial-statistics) |\n| Spatial domain | Contiguous tissue region of homogeneous expression, containing many types | a region | domain segmentation (this skill) |\n\nA domain can contain several niches; a niche can span domain boundaries. BANKSY and BASS switch between cell-typing and domain detection via a single parameter, which underscores that these are different outputs of related machinery, not the same task.\n\n## Domain Methods by Mechanism\n\nThe mechanism for \"using the neighbors\" is the axis that separates the methods. No method is a universal winner (Yuan 2024 *Nat Methods*); selection is scenario-specific.\n\n| Method | Spatial mechanism | Needs k? | Best when | Fails when |\n|--------|-------------------|----------|-----------|------------|\n| BayesSpace | MRF (Potts) smoothing prior on a t-mixture in PCA space; also enhances sub-spot resolution | yes (q) | Visium laminar\u002Fcontinuous tissue; want sub-spot enhancement | non-contiguous regions; MCMC is slow; smoothing strength is a fixed prior |\n| BASS | Bayesian hierarchical: cell-type AND domain jointly, Potts prior, multi-sample | yes (C, D) | joint cell-type + domain, multi-sample, low-continuity tissue | heavier to run; needs both counts set |\n| STAGATE | Graph attention autoencoder; learns per-edge weights, reconstructs from a spatially-smoothed latent | embedding free; downstream k for mclust | continuous tissue; attention down-weights cross-boundary edges (fights over-smoothing); scales to Slide-seq\u002FStereo-seq | black-box embedding; sensitive to graph radius |\n| GraphST | Graph self-supervised contrastive learning | yes (mclust) | low-res Visium; also does integration + deconvolution | sensitive to graph construction |\n| SpaGCN | GCN fusing expression + coordinates + histology RGB | searches resolution to hit target count | H&E histology is informative | needs registered histology |\n| SEDR | Masked autoencoder + variational graph autoencoder | yes (mclust) | Visium\u002FSlide-seq\u002FStereo-seq, robust on DLPFC | graph-construction sensitivity |\n| BANKSY | Neighbor-AUGMENTED features: own + neighborhood-mean + azimuthal Gabor; then Leiden\u002Fk-means | no fixed k | high-res imaging AND sequencing; unifies cell typing (lambda~0.2) and domains (lambda~0.8); very scalable; transparent | lambda mis-set collapses the task it solves |\n| UTAG | Message passing: multiply features by normalized adjacency (one-hop average), then Leiden | no fixed k | multiplexed imaging\u002Fproteomics (IMC, CODEX, MIBI); fast | one-hop smoothing only |\n| stLearn (SME) | Histology-CNN-weighted smoothing of each spot's expression, then clustering | downstream Louvain\u002Fk-means | H&E available and well-registered | only as good as image registration |\n\nMechanism families: MRF\u002FBayesian smoothing (BayesSpace, BASS, PRECAST) vs graph neural net (STAGATE, GraphST, SpaGCN, SEDR) vs neighbor-augmentation (BANKSY, UTAG) vs histology-guided (stLearn, SpaGCN). Benchmarks evolve fast -- verify the current verdict for the tissue geometry before committing.\n\n## Read the Benchmark With the DLPFC Caveat\n\nThe standard ground truth is DLPFC (Maynard 2021 *Nat Neurosci* 24:425-436): 12 Visium sections of human dorsolateral prefrontal cortex, manually annotated into 6 cortical layers + white matter, scored by ARI. The Yuan 2024 (*Nat Methods* 21:712-722) benchmark of 13 methods x 34 datasets concludes there is NO single winner -- methods are complementary across accuracy, spatial continuity, marker detection, scalability, and robustness. On DLPFC, GNN\u002Fgraph methods (STAGATE, SEDR, DeepST) and BayesSpace are among the most robust, and a technology-stratified benchmark (Chen 2025 *iMeta* 4:e70084) finds STAGATE\u002FGraphST best on low-resolution Visium while BASS\u002FstLearn\u002FBANKSY lead on high-resolution platforms.\n\nThe caveat: DLPFC is a CONTINUOUS, LAMINAR tissue, which flatters smoothing-friendly methods. Do not over-generalize these rankings to non-laminar tissue. ALL methods struggle with NON-CONTIGUOUS domains (the same region appearing in separated patches -- scattered tumor nests, immune aggregates) because the spatial prior assumes contiguity; for non-contiguous biology, lower the spatial weight or switch to a niche\u002Fneighborhood analysis rather than domain segmentation.\n\n## Build the Spatial Graph\n\n**Goal:** Construct the spatial neighbor graph that every domain method inherits.\n\n**Approach:** Use a hex lattice for Visium (6 neighbors) and a generic kNN\u002FDelaunay graph for imaging point clouds; the graph radius IS a spatial-weight knob (too dense over-smooths, too sparse fragments).\n\n```python\nimport squidpy as sq\nimport scanpy as sc\n\nadata = sc.read_h5ad('preprocessed.h5ad')\n\n# Visium hex lattice: 6 immediate neighbors. For imaging point clouds use\n# coord_type='generic' with KNN or Delaunay. Build in microns where possible --\n# a graph built in pixels has a different radius than one built in microns.\nsq.gr.spatial_neighbors(adata, coord_type='grid', n_neighs=6)\n```\n\n## Expression-Only Clustering Is the Salt-and-Pepper Baseline\n\n**Goal:** Show why a domain method is needed at all.\n\n**Approach:** Cluster on expression PCA with no spatial term; the result is spatially incoherent and demonstrates the failure domain methods correct.\n\n```python\nsc.pp.neighbors(adata, n_neighbors=15, n_pcs=30)\nsc.tl.leiden(adata, resolution=0.5, key_added='expr_leiden',\n             flavor='igraph', n_iterations=2, directed=False)\n\n# Plot on tissue: speckled, no contiguous regions -- this is the baseline a\n# domain method is built to beat, not a domain result.\nsq.pl.spatial_scatter(adata, color='expr_leiden')\n```\n\n## Neighbor-Augmented Domains (BANKSY-style) and the Spatial-Weight Knob\n\n**Goal:** Produce spatially coherent domains and make the over-smoothing knob explicit.\n\n**Approach:** BANKSY concatenates each cell's own expression with its neighborhood-mean expression, mixed by lambda; lambda~0.2 yields cell typing, lambda~0.8 yields domains. A transparent neighbor-augmented matrix reproduces the idea with Squidpy when the BANKSY package is unavailable -- the lambda here is the load-bearing decision, not the clustering call.\n\n```python\nimport numpy as np\nfrom sklearn.preprocessing import normalize\n\n# Mean expression over each spot's spatial neighbors (the neighborhood signal).\nW = normalize(adata.obsp['spatial_connectivities'], norm='l1', axis=1)\nneighbor_mean = W @ adata.obsm['X_pca']\n\n# lambda is the spatial-weight knob: ~0.8 for domains, ~0.2 for cell typing.\n# Too high -> blobs (boundaries erased); too low -> salt-and-pepper.\nlam = 0.8\naugmented = np.concatenate(\n    [np.sqrt(1 - lam) * adata.obsm['X_pca'], np.sqrt(lam) * neighbor_mean], axis=1)\nadata.obsm['X_banksy'] = augmented\n\nsc.pp.neighbors(adata, use_rep='X_banksy', key_added='banksy')\nsc.tl.leiden(adata, resolution=0.5, key_added='domains', neighbors_key='banksy',\n             flavor='igraph', n_iterations=2, directed=False)\nsq.pl.spatial_scatter(adata, color='domains')\n```\n\nIf the BANKSY package is installed, prefer it: `import banksy_py` (Python) or the R `Banksy` Bioconductor package compute the own + neighborhood-mean + azimuthal Gabor (AGF) features and expose lambda directly.\n\n## Sweep the Spatial Weight to Find the Over-Smoothing Edge\n\n**Goal:** Locate the lambda where boundaries are sharp but regions stay contiguous.\n\n**Approach:** Recompute domains across a lambda grid and inspect boundaries against histology; the right lambda is the largest value before distinct regions merge into blobs, judged visually, not by an internal score.\n\n```python\nfor lam in [0.2, 0.5, 0.8]:\n    aug = np.concatenate(\n        [np.sqrt(1 - lam) * adata.obsm['X_pca'],\n         np.sqrt(lam) * (W @ adata.obsm['X_pca'])], axis=1)\n    adata.obsm['X_banksy'] = aug\n    sc.pp.neighbors(adata, use_rep='X_banksy', key_added='banksy')\n    sc.tl.leiden(adata, resolution=0.5, key_added=f'domains_l{lam}',\n                 neighbors_key='banksy', flavor='igraph', n_iterations=2, directed=False)\n\n# Low lambda -> speckled; high lambda -> over-smoothed blobs. Pick by eye\n# against known tissue architecture, not by silhouette.\nsq.pl.spatial_scatter(adata, color=['domains_l0.2', 'domains_l0.5', 'domains_l0.8'])\n```\n\n## Choose k as a Biological Decision, With k+-1 Sensitivity\n\n**Goal:** Set the number of domains to the regionalization the question needs and show the answer is not fragile to it.\n\n**Approach:** When a method takes k directly (BayesSpace q, mclust on a GNN embedding), run k, k-1, k+1 and report all three; do not silently optimize k against a clustering score, which has no biological ground truth.\n\n```python\nfrom sklearn.mixture import GaussianMixture\n\nfor k in [6, 7, 8]:  # DLPFC has 6 layers + white matter -> k near 7\n    gm = GaussianMixture(n_components=k, covariance_type='full', random_state=0)\n    adata.obs[f'domains_k{k}'] = gm.fit_predict(adata.obsm['X_banksy']).astype(str)\n\n# Report k+-1 side by side; a domain that only appears at one k is a weak claim.\nsq.pl.spatial_scatter(adata, color=['domains_k6', 'domains_k7', 'domains_k8'])\n```\n\n## BayesSpace for Laminar Tissue (R)\n\n**Goal:** Apply an MRF smoothing prior, the robust choice on continuous Visium tissue.\n\n**Approach:** Preprocess, then `spatialCluster` with q domains; the smoothing prior couples neighboring spots. Run in R and import the labels.\n\n```r\nlibrary(BayesSpace)\n\nsce \u003C- readRDS('sce.rds')\nsce \u003C- spatialPreprocess(sce, platform = 'Visium', n.PCs = 15)\n# q is the biological k; nrep is MCMC iterations. spatialEnhance() can split\n# spots into subspots for higher resolution after spatialCluster().\nsce \u003C- spatialCluster(sce, q = 7, nrep = 10000)\nwrite.csv(data.frame(barcode = colnames(sce), domain = sce$spatial.cluster),\n          'bayesspace_domains.csv', row.names = FALSE)\n```\n\n## STAGATE for High-Resolution or Large Sections (optional)\n\n**Goal:** Learn a spatially-aware embedding whose attention down-weights cross-boundary edges.\n\n**Approach:** Build the STAGATE radius graph, train the graph-attention autoencoder, then cluster the embedding; set the radius to the over-smoothing knob.\n\n```python\nimport STAGATE  # optional dependency; pip install STAGATE_pyG or use the TF build\n\nSTAGATE.Cal_Spatial_Net(adata, rad_cutoff=150)  # rad_cutoff is the graph radius knob\nSTAGATE.Stats_Spatial_Net(adata)\nadata = STAGATE.train_STAGATE(adata)\n\nsc.pp.neighbors(adata, use_rep='STAGATE')\nsc.tl.leiden(adata, resolution=0.5, key_added='stagate_domains',\n             flavor='igraph', n_iterations=2, directed=False)\n```\n\n## Name Domains From Markers\n\n**Goal:** Attach anatomical labels to spatially coherent clusters.\n\n**Approach:** Rank per-domain markers, then map cluster IDs to region names; marker p-values from the same data that defined the domains are for ranking and labeling only, not inference (the double-dipping caveat from single-cell\u002Fclustering).\n\n```python\nsc.tl.rank_genes_groups(adata, groupby='domains', method='wilcoxon')\nmarkers = sc.get.rank_genes_groups_df(adata, group=None)\nprint(markers.groupby('group').head(5))\n\ndomain_names = {'0': 'White matter', '1': 'Layer 1', '2': 'Layer 2\u002F3'}\nadata.obs['region'] = adata.obs['domains'].map(domain_names)\nsq.pl.spatial_scatter(adata, color='region')\n```\n\n## Common Errors\n\n| Symptom | Cause | Fix |\n|---------|-------|-----|\n| Speckled, spatially incoherent labels | No spatial term (plain Leiden on expression) or spatial weight too low | Use a domain method; raise lambda \u002F smoothing \u002F graph density |\n| Regions merged into smooth blobs, boundaries gone | Over-smoothing -- spatial weight too high or graph radius too large | Lower lambda \u002F smoothing strength \u002F rad_cutoff; check boundaries against histology |\n| Scattered tumor nests collapse into one domain or vanish | Non-contiguous domain; spatial prior assumes contiguity (Yuan 2024) | Lower the spatial weight, or switch to niche\u002Fneighborhood analysis (spatial-statistics) |\n| Domain count feels arbitrary \u002F reviewer questions k | k optimized against a silhouette score instead of chosen biologically | Set k from the biology; report k+-1 sensitivity; justify the regionalization |\n| Domains track a single sample\u002Fsection | Batch confounded with biology in multi-sample data | Use a multi-sample method (BASS, PRECAST, GraphST integration) or integrate first |\n| Answer changes with no parameter change | Spatial graph rebuilt with different units (pixels vs microns) or different kNN\u002FDelaunay | Pin the graph construction and coordinate units; build once, reuse |\n| \"Domain\" answer to a \"which types co-occur\" question | Domain segmentation used for a niche question | Use neighborhood enrichment \u002F co-occurrence (spatial-statistics) instead |\n| Marker p-values quoted as proof a domain is real | Double-dipping (testing the clustering that defined the groups) | Use markers for ranking\u002Flabeling only; validate regions against known architecture |\n\n## Related Skills\n\n- spatial-neighbors - Build and tune the spatial graph every domain method inherits\n- spatial-statistics - Niche\u002Fneighborhood enrichment and co-occurrence when the question is which cell types co-occur, not which region this is\n- spatial-deconvolution - Per-spot cell-type composition; domains are regions, deconvolution is composition within a spot\n- single-cell\u002Fclustering - Non-spatial clustering, resolution sweeps, and the double-dipping caveat on post-clustering marker tests\n\n## References\n\n- Zhao et al. (2021). Spatial transcriptomics at subspot resolution with BayesSpace. Nat Biotechnol 39:1375-1384.\n- Hu et al. (2021). SpaGCN: integrating gene expression, spatial location and histology to identify spatial domains and SVGs by graph convolutional network. Nat Methods 18:1342-1351.\n- Dong & Zhang (2022). Deciphering spatial domains from spatially resolved transcriptomics with an adaptive graph attention auto-encoder (STAGATE). Nat Commun 13:1739.\n- Long et al. (2023). Spatially informed clustering, integration, and deconvolution of spatial transcriptomics with GraphST. Nat Commun 14:1155.\n- Singhal et al. (2024). BANKSY unifies cell typing and tissue domain segmentation for scalable spatial omics data analysis. Nat Genet 56:431-441.\n- Kim et al. (2022). Unsupervised discovery of tissue architecture in multiplexed imaging (UTAG). Nat Methods 19:1653-1661.\n- Xu et al. (2024). Unsupervised spatially embedded deep representation of spatial transcriptomics (SEDR). Genome Med 16:12.\n- Li & Zhou (2022). BASS: multi-scale and multi-sample analysis enables accurate cell type clustering and spatial domain detection in spatial transcriptomic studies. Genome Biol 23:168.\n- Yuan et al. (2024). Benchmarking spatial clustering methods with spatially resolved transcriptomics data. Nat Methods 21:712-722.\n- Chen et al. (2025). A comprehensive benchmarking for spatially resolved transcriptomics clustering methods across variable technologies, organs, and replicates. iMeta 4:e70084.\n- Maynard et al. (2021). Transcriptome-scale spatial gene expression in the human dorsolateral prefrontal cortex. Nat Neurosci 24:425-436.\n- Palla et al. (2022). Squidpy: a scalable framework for spatial omics analysis. Nat Methods 19:171-178.\n","## Version Compatibility\n\nReference examples tested with: squidpy 1.4+, scanpy 1.10+, anndata 0.10+, scikit-learn 1.4+; BANKSY (banksy_py \u002F R Banksy), BayesSpace 1.12+ (R), STAGATE\u002FGraphST optional\n\nBefore using code patterns, verify installed versions match. If versions differ:\n- Python: `pip show \u003Cpackage>` then `help(module.function)` to check signatures\n- R: `packageVersion('\u003Cpkg>')` then `?function_name` to verify parameters\n\nIf code throws ImportError, AttributeError, or TypeError, introspect the installed\npackage and adapt the example to match the actual API rather than retrying.\n\n# Spatial Domain Detection\n\n**\"Identify tissue domains in my section\"** -> Partition the tissue into spatially contiguous regions of homogeneous expression\u002Fcomposition, using BOTH transcriptional similarity AND spatial proximity, so the output is a region label per spot\u002Fcell that is spatially coherent.\n- Python: `squidpy.gr.spatial_neighbors` for the graph, then a neighbor-augmented or graph-based domain method (BANKSY, STAGATE, GraphST)\n- R: BayesSpace (`spatialCluster`), BASS, or Banksy\n\n## Governing Principle\n\nA spatial domain is a REGION, not a cell type, and not a niche -- conflating the three is the central conceptual error of this analysis. A CELL TYPE is one cell's transcriptional identity. A NICHE (cellular neighborhood) is the local cell-type COMPOSITION around a cell -- which types co-occur. A SPATIAL DOMAIN is a contiguous tissue region (a cortical layer, tumor core, stromal band) that contains MANY cell types and several niches. If the question is \"which cell types co-occur,\" that is a niche question and belongs to neighborhood-enrichment analysis (spatial-statistics), NOT domain segmentation; running the wrong one answers the wrong question.\n\nDomain methods exist because plain Leiden\u002FLouvain on expression alone ignores coordinates and produces salt-and-pepper, spatially-incoherent labels. Every domain method deliberately adds a spatial term so neighbors tend to share a label. The load-bearing decision is therefore NOT the clustering algorithm -- it is the SPATIAL-WEIGHT knob (BANKSY lambda, BayesSpace smoothing, SpaGCN histology weight, GNN graph radius). Too much spatial weight is the #1 trap, over-smoothing: it erases real boundaries and merges biologically distinct regions into blobs. Too little reverts to salt-and-pepper. The number of domains k is the second decision, and it is a BIOLOGICAL choice (how fine a regionalization the question needs), not a statistic to optimize against a silhouette score -- report k, justify it, and show sensitivity to k+-1.\n\n## Domain vs Niche vs Cell Type\n\n| Concept | What it is | Unit | Right tool |\n|---------|-----------|------|-----------|\n| Cell type | One cell's transcriptional identity (lineage\u002Fstate) | a cell | clustering + markers (single-cell\u002Fclustering) |\n| Niche \u002F cellular neighborhood | Local cell-type composition -- which types co-occur around a cell | a neighborhood | neighborhood enrichment \u002F co-occurrence (spatial-statistics) |\n| Spatial domain | Contiguous tissue region of homogeneous expression, containing many types | a region | domain segmentation (this skill) |\n\nA domain can contain several niches; a niche can span domain boundaries. BANKSY and BASS switch between cell-typing and domain detection via a single parameter, which underscores that these are different outputs of related machinery, not the same task.\n\n## Domain Methods by Mechanism\n\nThe mechanism for \"using the neighbors\" is the axis that separates the methods. No method is a universal winner (Yuan 2024 *Nat Methods*); selection is scenario-specific.\n\n| Method | Spatial mechanism | Needs k? | Best when | Fails when |\n|--------|-------------------|----------|-----------|------------|\n| BayesSpace | MRF (Potts) smoothing prior on a t-mixture in PCA space; also enhances sub-spot resolution | yes (q) | Visium laminar\u002Fcontinuous tissue; want sub-spot enhancement | non-contiguous regions; MCMC is slow; smoothing strength is a fixed prior |\n| BASS | Bayesian hierarchical: cell-type AND domain jointly, Potts prior, multi-sample | yes (C, D) | joint cell-type + domain, multi-sample, low-continuity tissue | heavier to run; needs both counts set |\n| STAGATE | Graph attention autoencoder; learns per-edge weights, reconstructs from a spatially-smoothed latent | embedding free; downstream k for mclust | continuous tissue; attention down-weights cross-boundary edges (fights over-smoothing); scales to Slide-seq\u002FStereo-seq | black-box embedding; sensitive to graph radius |\n| GraphST | Graph self-supervised contrastive learning | yes (mclust) | low-res Visium; also does integration + deconvolution | sensitive to graph construction |\n| SpaGCN | GCN fusing expression + coordinates + histology RGB | searches resolution to hit target count | H&E histology is informative | needs registered histology |\n| SEDR | Masked autoencoder + variational graph autoencoder | yes (mclust) | Visium\u002FSlide-seq\u002FStereo-seq, robust on DLPFC | graph-construction sensitivity |\n| BANKSY | Neighbor-AUGMENTED features: own + neighborhood-mean + azimuthal Gabor; then Leiden\u002Fk-means | no fixed k | high-res imaging AND sequencing; unifies cell typing (lambda~0.2) and domains (lambda~0.8); very scalable; transparent | lambda mis-set collapses the task it solves |\n| UTAG | Message passing: multiply features by normalized adjacency (one-hop average), then Leiden | no fixed k | multiplexed imaging\u002Fproteomics (IMC, CODEX, MIBI); fast | one-hop smoothing only |\n| stLearn (SME) | Histology-CNN-weighted smoothing of each spot's expression, then clustering | downstream Louvain\u002Fk-means | H&E available and well-registered | only as good as image registration |\n\nMechanism families: MRF\u002FBayesian smoothing (BayesSpace, BASS, PRECAST) vs graph neural net (STAGATE, GraphST, SpaGCN, SEDR) vs neighbor-augmentation (BANKSY, UTAG) vs histology-guided (stLearn, SpaGCN). Benchmarks evolve fast -- verify the current verdict for the tissue geometry before committing.\n\n## Read the Benchmark With the DLPFC Caveat\n\nThe standard ground truth is DLPFC (Maynard 2021 *Nat Neurosci* 24:425-436): 12 Visium sections of human dorsolateral prefrontal cortex, manually annotated into 6 cortical layers + white matter, scored by ARI. The Yuan 2024 (*Nat Methods* 21:712-722) benchmark of 13 methods x 34 datasets concludes there is NO single winner -- methods are complementary across accuracy, spatial continuity, marker detection, scalability, and robustness. On DLPFC, GNN\u002Fgraph methods (STAGATE, SEDR, DeepST) and BayesSpace are among the most robust, and a technology-stratified benchmark (Chen 2025 *iMeta* 4:e70084) finds STAGATE\u002FGraphST best on low-resolution Visium while BASS\u002FstLearn\u002FBANKSY lead on high-resolution platforms.\n\nThe caveat: DLPFC is a CONTINUOUS, LAMINAR tissue, which flatters smoothing-friendly methods. Do not over-generalize these rankings to non-laminar tissue. ALL methods struggle with NON-CONTIGUOUS domains (the same region appearing in separated patches -- scattered tumor nests, immune aggregates) because the spatial prior assumes contiguity; for non-contiguous biology, lower the spatial weight or switch to a niche\u002Fneighborhood analysis rather than domain segmentation.\n\n## Build the Spatial Graph\n\n**Goal:** Construct the spatial neighbor graph that every domain method inherits.\n\n**Approach:** Use a hex lattice for Visium (6 neighbors) and a generic kNN\u002FDelaunay graph for imaging point clouds; the graph radius IS a spatial-weight knob (too dense over-smooths, too sparse fragments).\n\n```python\nimport squidpy as sq\nimport scanpy as sc\n\nadata = sc.read_h5ad('preprocessed.h5ad')\n\n# Visium hex lattice: 6 immediate neighbors. For imaging point clouds use\n# coord_type='generic' with KNN or Delaunay. Build in microns where possible --\n# a graph built in pixels has a different radius than one built in microns.\nsq.gr.spatial_neighbors(adata, coord_type='grid', n_neighs=6)\n```\n\n## Expression-Only Clustering Is the Salt-and-Pepper Baseline\n\n**Goal:** Show why a domain method is needed at all.\n\n**Approach:** Cluster on expression PCA with no spatial term; the result is spatially incoherent and demonstrates the failure domain methods correct.\n\n```python\nsc.pp.neighbors(adata, n_neighbors=15, n_pcs=30)\nsc.tl.leiden(adata, resolution=0.5, key_added='expr_leiden',\n             flavor='igraph', n_iterations=2, directed=False)\n\n# Plot on tissue: speckled, no contiguous regions -- this is the baseline a\n# domain method is built to beat, not a domain result.\nsq.pl.spatial_scatter(adata, color='expr_leiden')\n```\n\n## Neighbor-Augmented Domains (BANKSY-style) and the Spatial-Weight Knob\n\n**Goal:** Produce spatially coherent domains and make the over-smoothing knob explicit.\n\n**Approach:** BANKSY concatenates each cell's own expression with its neighborhood-mean expression, mixed by lambda; lambda~0.2 yields cell typing, lambda~0.8 yields domains. A transparent neighbor-augmented matrix reproduces the idea with Squidpy when the BANKSY package is unavailable -- the lambda here is the load-bearing decision, not the clustering call.\n\n```python\nimport numpy as np\nfrom sklearn.preprocessing import normalize\n\n# Mean expression over each spot's spatial neighbors (the neighborhood signal).\nW = normalize(adata.obsp['spatial_connectivities'], norm='l1', axis=1)\nneighbor_mean = W @ adata.obsm['X_pca']\n\n# lambda is the spatial-weight knob: ~0.8 for domains, ~0.2 for cell typing.\n# Too high -> blobs (boundaries erased); too low -> salt-and-pepper.\nlam = 0.8\naugmented = np.concatenate(\n    [np.sqrt(1 - lam) * adata.obsm['X_pca'], np.sqrt(lam) * neighbor_mean], axis=1)\nadata.obsm['X_banksy'] = augmented\n\nsc.pp.neighbors(adata, use_rep='X_banksy', key_added='banksy')\nsc.tl.leiden(adata, resolution=0.5, key_added='domains', neighbors_key='banksy',\n             flavor='igraph', n_iterations=2, directed=False)\nsq.pl.spatial_scatter(adata, color='domains')\n```\n\nIf the BANKSY package is installed, prefer it: `import banksy_py` (Python) or the R `Banksy` Bioconductor package compute the own + neighborhood-mean + azimuthal Gabor (AGF) features and expose lambda directly.\n\n## Sweep the Spatial Weight to Find the Over-Smoothing Edge\n\n**Goal:** Locate the lambda where boundaries are sharp but regions stay contiguous.\n\n**Approach:** Recompute domains across a lambda grid and inspect boundaries against histology; the right lambda is the largest value before distinct regions merge into blobs, judged visually, not by an internal score.\n\n```python\nfor lam in [0.2, 0.5, 0.8]:\n    aug = np.concatenate(\n        [np.sqrt(1 - lam) * adata.obsm['X_pca'],\n         np.sqrt(lam) * (W @ adata.obsm['X_pca'])], axis=1)\n    adata.obsm['X_banksy'] = aug\n    sc.pp.neighbors(adata, use_rep='X_banksy', key_added='banksy')\n    sc.tl.leiden(adata, resolution=0.5, key_added=f'domains_l{lam}',\n                 neighbors_key='banksy', flavor='igraph', n_iterations=2, directed=False)\n\n# Low lambda -> speckled; high lambda -> over-smoothed blobs. Pick by eye\n# against known tissue architecture, not by silhouette.\nsq.pl.spatial_scatter(adata, color=['domains_l0.2', 'domains_l0.5', 'domains_l0.8'])\n```\n\n## Choose k as a Biological Decision, With k+-1 Sensitivity\n\n**Goal:** Set the number of domains to the regionalization the question needs and show the answer is not fragile to it.\n\n**Approach:** When a method takes k directly (BayesSpace q, mclust on a GNN embedding), run k, k-1, k+1 and report all three; do not silently optimize k against a clustering score, which has no biological ground truth.\n\n```python\nfrom sklearn.mixture import GaussianMixture\n\nfor k in [6, 7, 8]:  # DLPFC has 6 layers + white matter -> k near 7\n    gm = GaussianMixture(n_components=k, covariance_type='full', random_state=0)\n    adata.obs[f'domains_k{k}'] = gm.fit_predict(adata.obsm['X_banksy']).astype(str)\n\n# Report k+-1 side by side; a domain that only appears at one k is a weak claim.\nsq.pl.spatial_scatter(adata, color=['domains_k6', 'domains_k7', 'domains_k8'])\n```\n\n## BayesSpace for Laminar Tissue (R)\n\n**Goal:** Apply an MRF smoothing prior, the robust choice on continuous Visium tissue.\n\n**Approach:** Preprocess, then `spatialCluster` with q domains; the smoothing prior couples neighboring spots. Run in R and import the labels.\n\n```r\nlibrary(BayesSpace)\n\nsce \u003C- readRDS('sce.rds')\nsce \u003C- spatialPreprocess(sce, platform = 'Visium', n.PCs = 15)\n# q is the biological k; nrep is MCMC iterations. spatialEnhance() can split\n# spots into subspots for higher resolution after spatialCluster().\nsce \u003C- spatialCluster(sce, q = 7, nrep = 10000)\nwrite.csv(data.frame(barcode = colnames(sce), domain = sce$spatial.cluster),\n          'bayesspace_domains.csv', row.names = FALSE)\n```\n\n## STAGATE for High-Resolution or Large Sections (optional)\n\n**Goal:** Learn a spatially-aware embedding whose attention down-weights cross-boundary edges.\n\n**Approach:** Build the STAGATE radius graph, train the graph-attention autoencoder, then cluster the embedding; set the radius to the over-smoothing knob.\n\n```python\nimport STAGATE  # optional dependency; pip install STAGATE_pyG or use the TF build\n\nSTAGATE.Cal_Spatial_Net(adata, rad_cutoff=150)  # rad_cutoff is the graph radius knob\nSTAGATE.Stats_Spatial_Net(adata)\nadata = STAGATE.train_STAGATE(adata)\n\nsc.pp.neighbors(adata, use_rep='STAGATE')\nsc.tl.leiden(adata, resolution=0.5, key_added='stagate_domains',\n             flavor='igraph', n_iterations=2, directed=False)\n```\n\n## Name Domains From Markers\n\n**Goal:** Attach anatomical labels to spatially coherent clusters.\n\n**Approach:** Rank per-domain markers, then map cluster IDs to region names; marker p-values from the same data that defined the domains are for ranking and labeling only, not inference (the double-dipping caveat from single-cell\u002Fclustering).\n\n```python\nsc.tl.rank_genes_groups(adata, groupby='domains', method='wilcoxon')\nmarkers = sc.get.rank_genes_groups_df(adata, group=None)\nprint(markers.groupby('group').head(5))\n\ndomain_names = {'0': 'White matter', '1': 'Layer 1', '2': 'Layer 2\u002F3'}\nadata.obs['region'] = adata.obs['domains'].map(domain_names)\nsq.pl.spatial_scatter(adata, color='region')\n```\n\n## Common Errors\n\n| Symptom | Cause | Fix |\n|---------|-------|-----|\n| Speckled, spatially incoherent labels | No spatial term (plain Leiden on expression) or spatial weight too low | Use a domain method; raise lambda \u002F smoothing \u002F graph density |\n| Regions merged into smooth blobs, boundaries gone | Over-smoothing -- spatial weight too high or graph radius too large | Lower lambda \u002F smoothing strength \u002F rad_cutoff; check boundaries against histology |\n| Scattered tumor nests collapse into one domain or vanish | Non-contiguous domain; spatial prior assumes contiguity (Yuan 2024) | Lower the spatial weight, or switch to niche\u002Fneighborhood analysis (spatial-statistics) |\n| Domain count feels arbitrary \u002F reviewer questions k | k optimized against a silhouette score instead of chosen biologically | Set k from the biology; report k+-1 sensitivity; justify the regionalization |\n| Domains track a single sample\u002Fsection | Batch confounded with biology in multi-sample data | Use a multi-sample method (BASS, PRECAST, GraphST integration) or integrate first |\n| Answer changes with no parameter change | Spatial graph rebuilt with different units (pixels vs microns) or different kNN\u002FDelaunay | Pin the graph construction and coordinate units; build once, reuse |\n| \"Domain\" answer to a \"which types co-occur\" question | Domain segmentation used for a niche question | Use neighborhood enrichment \u002F co-occurrence (spatial-statistics) instead |\n| Marker p-values quoted as proof a domain is real | Double-dipping (testing the clustering that defined the groups) | Use markers for ranking\u002Flabeling only; validate regions against known architecture |\n\n## Related Skills\n\n- spatial-neighbors - Build and tune the spatial graph every domain method inherits\n- spatial-statistics - Niche\u002Fneighborhood enrichment and co-occurrence when the question is which cell types co-occur, not which region this is\n- spatial-deconvolution - Per-spot cell-type composition; domains are regions, deconvolution is composition within a spot\n- single-cell\u002Fclustering - Non-spatial clustering, resolution sweeps, and the double-dipping caveat on post-clustering marker tests\n\n## References\n\n- Zhao et al. (2021). Spatial transcriptomics at subspot resolution with BayesSpace. Nat Biotechnol 39:1375-1384.\n- Hu et al. (2021). SpaGCN: integrating gene expression, spatial location and histology to identify spatial domains and SVGs by graph convolutional network. Nat Methods 18:1342-1351.\n- Dong & Zhang (2022). Deciphering spatial domains from spatially resolved transcriptomics with an adaptive graph attention auto-encoder (STAGATE). Nat Commun 13:1739.\n- Long et al. (2023). Spatially informed clustering, integration, and deconvolution of spatial transcriptomics with GraphST. Nat Commun 14:1155.\n- Singhal et al. (2024). BANKSY unifies cell typing and tissue domain segmentation for scalable spatial omics data analysis. Nat Genet 56:431-441.\n- Kim et al. (2022). Unsupervised discovery of tissue architecture in multiplexed imaging (UTAG). Nat Methods 19:1653-1661.\n- Xu et al. (2024). Unsupervised spatially embedded deep representation of spatial transcriptomics (SEDR). Genome Med 16:12.\n- Li & Zhou (2022). BASS: multi-scale and multi-sample analysis enables accurate cell type clustering and spatial domain detection in spatial transcriptomic studies. Genome Biol 23:168.\n- Yuan et al. (2024). Benchmarking spatial clustering methods with spatially resolved transcriptomics data. Nat Methods 21:712-722.\n- Chen et al. (2025). A comprehensive benchmarking for spatially resolved transcriptomics clustering methods across variable technologies, organs, and replicates. iMeta 4:e70084.\n- Maynard et al. (2021). Transcriptome-scale spatial gene expression in the human dorsolateral prefrontal cortex. Nat Neurosci 24:425-436.\n- Palla et al. (2022). Squidpy: a scalable framework for spatial omics analysis. Nat Methods 19:171-178.\n",{"name":5,"source":24,"license":15,"tool_type":25,"description":7,"imported_at":26,"source_path":27,"primary_tool":28,"source_repository":29},"https:\u002F\u002Fgithub.com\u002FGPTomics\u002FbioSkills","python","2026-09-20","spatial-transcriptomics\u002Fspatial-domains\u002FSKILL.md","squidpy","GPTomics\u002FbioSkills","main","2d1750717747097b602acf30bd09a7cbd4b4c6ea","awesome_skills\u002Fbiology\u002Fbio-spatial-transcriptomics-spatial-domains","https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous\u002FAll-skills","https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous","2026-11-09 11:45:36",{"source":8,"skillName":5,"commands":37,"prompt":47,"zipAvailable":48},[38,41,44],{"tool":39,"command":40},"npx","npx skills add Mr-Nobody-Anonymous\u002FAll-skills --skill bio-spatial-transcriptomics-spatial-domains",{"tool":42,"command":43},"bunx","bunx skills add Mr-Nobody-Anonymous\u002FAll-skills --skill bio-spatial-transcriptomics-spatial-domains",{"tool":45,"command":46},"pnpm","pnpm dlx skills add Mr-Nobody-Anonymous\u002FAll-skills --skill bio-spatial-transcriptomics-spatial-domains","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous\u002FAll-skills\u002Fblob\u002Fmain\u002Fawesome_skills\u002Fbiology\u002Fbio-spatial-transcriptomics-spatial-domains\u002FSKILL.md",true,[50,56],{"path":51,"resourceType":52,"resourceTypeDesc":53,"sizeBytes":54,"sourceHtmlUrl":55},"examples\u002Fdetect_domains.py","SCRIPT","脚本",2686,"https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous\u002FAll-skills\u002Fblob\u002Fmain\u002Fawesome_skills\u002Fbiology\u002Fbio-spatial-transcriptomics-spatial-domains\u002Fexamples\u002Fdetect_domains.py",{"path":57,"resourceType":58,"resourceTypeDesc":59,"sizeBytes":60,"sourceHtmlUrl":61},"usage-guide.md","REFERENCE","参考文档",4493,"https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous\u002FAll-skills\u002Fblob\u002Fmain\u002Fawesome_skills\u002Fbiology\u002Fbio-spatial-transcriptomics-spatial-domains\u002Fusage-guide.md",[63,71,78,85,92,99],{"skillNo":64,"name":65,"nameZh":6,"description":66,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":67,"categories":68,"keywords":69,"publishedAt":70},"01M7FFFZPC2025JQQHPHDX154J","dei-assessment","Assess diversity, equity, and inclusion with metrics. TRIGGERS - Use when user needs help with dei-assessment related tasks.","https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous\u002FAll-skills\u002Fblob\u002Fmain\u002Fawesome_skills\u002Fhr-people\u002Fdei-assessment\u002FSKILL.md",[],[],"2026-11-16 10:05:34",{"skillNo":72,"name":73,"nameZh":6,"description":74,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":75,"categories":76,"keywords":77,"publishedAt":70},"01M7FFFWN8H4WA7A6Y398QKB6J","dei-action-plan","Create DEI action plans with goals and initiatives. TRIGGERS - Use when user needs help with dei-action-plan related tasks.","https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous\u002FAll-skills\u002Fblob\u002Fmain\u002Fawesome_skills\u002Fhr-people\u002Fdei-action-plan\u002FSKILL.md",[],[],{"skillNo":79,"name":80,"nameZh":6,"description":81,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":82,"categories":83,"keywords":84,"publishedAt":70},"01M7FFFSHTM0QZRVZ2NRJTJ0PY","culture-fit-assessment","Design culture add assessment processes. TRIGGERS - Use when user needs help with culture-fit-assessment related tasks.","https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous\u002FAll-skills\u002Fblob\u002Fmain\u002Fawesome_skills\u002Fhr-people\u002Fculture-fit-assessment\u002FSKILL.md",[],[],{"skillNo":86,"name":87,"nameZh":6,"description":88,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":89,"categories":90,"keywords":91,"publishedAt":70},"01M7FFFMA12RKEQA38KHJSXXPS","cowork-to-code-bridge","Use an already-installed, independently verified cowork-to-code bridge to run narrowly approved actions on the user's own macOS, Linux, or WSL2 machine through a local file queue.","https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous\u002FAll-skills\u002Fblob\u002Fmain\u002Fawesome_skills\u002Fhr-people\u002Fcowork-to-code-bridge\u002FSKILL.md",[],[],{"skillNo":93,"name":94,"nameZh":6,"description":95,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":96,"categories":97,"keywords":98,"publishedAt":70},"01M7FFFH1F6M8YRGW5V487GH1K","counter-offer-response","Draft counter-offer responses for retention. TRIGGERS - Use when user needs help with counter-offer-response related tasks.","https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous\u002FAll-skills\u002Fblob\u002Fmain\u002Fawesome_skills\u002Fhr-people\u002Fcounter-offer-response\u002FSKILL.md",[],[],{"skillNo":100,"name":101,"nameZh":6,"description":102,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":103,"categories":104,"keywords":105,"publishedAt":70},"01M7FFFDGEHV8Y7J9N40HZXW52","continuous-feedback-system","Design continuous feedback systems. TRIGGERS - Use when user needs help with continuous-feedback-system related tasks.","https:\u002F\u002Fgithub.com\u002FMr-Nobody-Anonymous\u002FAll-skills\u002Fblob\u002Fmain\u002Fawesome_skills\u002Fhr-people\u002Fcontinuous-feedback-system\u002FSKILL.md",[],[]]