paga.json

{
  "tool_id": "paga",
  "name": "PAGA Trajectory Inference",
  "category": "trajectory",
  "framework": "scanpy",
  "function": "sc.tl.paga",
  "valid_after": ["clustering"],
  "valid_before": ["monocle3_pseudotime"],
  "paradigms": ["developmental_trajectory", "cell_atlas"],
  "parameters": {
    "groups": {
      "type": "string",
      "default": "leiden",
      "guidance": "obs key with cluster labels. PAGA computes connectivity between these groups."
    },
    "model": {
      "type": "string",
      "default": "v1.2",
      "options": ["v1.0", "v1.2"],
      "guidance": "PAGA model version. v1.2 is more robust."
    }
  },
  "outputs": {
    "connectivities": "adata.uns['paga']['connectivities']",
    "connectivities_tree": "adata.uns['paga']['connectivities_tree']"
  },
  "validation": {
    "check": "PAGA graph should be connected (all clusters reachable). Disconnected components may indicate over-clustering or truly distinct populations."
  },
  "notes": "PAGA provides a coarse-grained trajectory graph at the cluster level. Useful for identifying plausible lineage relationships before running pseudotime. Combine with sc.pl.paga for visualization and sc.tl.draw_graph for force-directed layouts initialized by PAGA.",
  "provenance_captures": ["groups", "model", "n_groups"]
}