chore: Updating claude plugin (#52)

* chore: Updating claude plugin

* improved assay ontology recommendation

---------

Co-authored-by: mvenkatakrishnan <mvenkatakrishnan@gmail.com>

Author: manasaV3

plugins/catalog/skills/catalog-register/SKILL.md

@@ -205,6 +205,30 @@ unset — empty is honest; a fabricated value is a data-quality bug.
   ask whether each `location` (data asset) should point at the next level down
   (e.g. one location per well/FOV) rather than the whole store. Map `locations`
   accordingly — don't assume one dataset = one `.zarr` root.
+11. **Derive `organism` / `tissue` / `disease` only when *formally entailed* by an
+  existing source value — confirm before coding, never hallucinate.** Sample
+  fields like organism, tissue and disease are frequently absent as their own column
+  yet fully determined by another column. A `cell_line` / `cell_type` pins down
+  both the species and the tissue of origin. You may fill such a field **only**
+  when the value is *derived*, never guessed.
+   - **Formal derivation, not intuition.** The target must be *entailed* by an
+     authority, not inferred by feel. The mechanism is the OLS hierarchy from
+     rule 9: resolve the source label (e.g. `cell_type="A549"` → `EFO:0001086`),
+     then read its ancestors. If an unambiguous ancestor pins the field
+     (`A549` → parent `Homo sapiens cell line` ⇒ `organism = Homo sapiens /
+     NCBITaxon:9606`).
+   - **Ambiguous ⇒ leave unset.** If the source value does not *uniquely*
+     determine the target — e.g. a cell *type* aligns with models in multiple species — do **not** derive. Leave it unset
+     and note it; never pick the "likely" option.
+   - **Confirm with the user before writing the derivation into code.**
+     Derivation is a mapping *decision*, not a lookup. Show the user the rule you intend to encode as source field → target field, the entailing authority (which ancestor/lookup), and
+     exactly which rows will stay unset because they're ambiguous. Get
+     explicit approval. Only then code it. Don't silently bake in a derivation.
+   - **Cell lines belong in `tissue`.** When you do derive, a cell line's
+     canonical slot is `sample.tissue = [TissueEntry(label=<line>,
+     ontology_id=<resolved id>, type="cell line")]` — not an ad-hoc `cell_line`
+     extra. Its anatomical tissue-of-origin (e.g. A549 → lung / UBERON) is a
+     *further* derivation, held to the same entailment + confirmation bar.
 - **`version` is never null.** It's a required signature field (`str`, not
   optional). If the source has no version, default it to `"1.0.0"` — use
   `src.get("release") or "1.0.0"`, never pass `None`.
@@ -244,7 +268,7 @@ Get token from catalog's `/docs` → Token → `/token/issue`. Pass via `CATALOG
   here with a precise error, before any network call.
 - **`record_schema_version` is auto-set.** Don't map a source field
   onto it; the model defaults it.
-- **`dataset_type` values are `raw` / `processed`** (`DatasetType.raw`) — not `primary`.
+- **`dataset_type` values are `raw` / `processed`** (`DatasetType.raw`)
 - **`data_quality.checks_*` accept any shape** (`Any`): a count, a list of
   names, or a nested dict all validate.
 - **`extra="allow"` cuts both ways.** Unknown keys are preserved (great for
@@ -262,6 +286,7 @@ Get token from catalog's `/docs` → Token → `/token/issue`. Pass via `CATALOG
 - **No live server on a clean machine.** `--dry-run` swaps the client's internal
   `_http` for an `httpx.MockTransport` fake catalog (`_mock_client` in the
   template) so you can validate the full submit path without a token.
+- Don't ask user for sensitive tokens or credentials.
 
 ## Troubleshooting
 
@@ -277,4 +302,4 @@ Get token from catalog's `/docs` → Token → `/token/issue`. Pass via `CATALOG
   to update it in place (see *Registering*).
 - `ValueError: update_if_exists and error_on_duplicate cannot both be True` →
   both flags were set; `update_if_exists=True` requires `error_on_duplicate=False`.
-- `AuthenticationError` (401) → bad/expired token; reissue at `/token/issue`.
+- `AuthenticationError` (401) → bad/expired token; ask user to reissue at `/token` page.

schema/v1.4.0/schema.md

@@ -115,15 +115,16 @@ and modality-specific metadata all live at the dataset level.
 
 ### Dataset types
 
-**Raw** are data produced by a single experimental acquisition — one pass through
+**Raw** are data produced by a single experimental acquisition from a machine — one pass through
 an instrument or pipeline under consistent conditions (same organism, same assay, same
 modality). If two groups of files came from the same run under the same conditions,
 they go in one dataset. If anything changes between them, register them separately.
 
 **Processed** are always new datasets. When an existing dataset is processed,
 the output goes in a new dataset record linked back to its source via a lineage edge.
-The original is never modified. A processed dataset can be derived from multiple source
-datasets across different samples, assays, or modalities.
+The original is never modified. Any qualitative change to the raw dataset such as preprocessing,
+makes it a processed dataset. A processed dataset can be derived from multiple source datasets
+across different samples, assays, or modalities.
 
 The `dataset_type` field records this distinction as `raw` or `processed`.
 
@@ -182,12 +183,12 @@ level are permitted for domain-specific or team-specific needs.
 
 #### Experimental metadata
 
-| Field | Type | Required | Description |
-|---|---|---|---|
-| `sub_modality` | string | No | More granular specification of the experimental procedure (e.g. `scRNA-seq`, `brightfield`, `bulk`). |
-| `assay` | list[json] | No | Assay(s) used to produce the dataset. Each entry: `{ label, ontology_id }`. Recommended ontology: **EFO** (e.g. `EFO:0022605`). |
-| `machine_information` | json | No | Information about the instrument used for data generation. |
-| `experimental_protocols` | json | No | Protocol details for the experiment. |
+| Field | Type | Required | Description                                                                                                                                                     |
+|---|---|---|-----------------------------------------------------------------------------------------------------------------------------------------------------------------|
+| `sub_modality` | string | No | More granular specification of the experimental procedure (e.g. `scRNA-seq`, `brightfield`, `bulk`).                                                            |
+| `assay` | list[json] | No | Assay(s) used to produce the dataset. Each entry: `{ label, ontology_id }`. Recommended ontology: **EFO** (e.g. `EFO:0022605`), **FBbi** (eg. `FBbi:00100015`). |
+| `machine_information` | json | No | Information about the instrument used for data generation.                                                                                                      |
+| `experimental_protocols` | json | No | Protocol details for the experiment.                                                                                                                            |
 
 #### Sample metadata
 
@@ -208,13 +209,13 @@ ontology recommended below. These follow the
 [CZI cross-modality standard](https://github.com/chanzuckerberg/data-guidance/blob/main/standards/cross-modality/1.1.0/schema.md);
 the `label` should be the ontology term's preferred label.
 
-| Field | Recommended ontology | Notes & special values |
-|---|---|---|
-| `organism` | **NCBITaxon** | e.g. `NCBITaxon:9606` (human), `NCBITaxon:10090` (mouse), `NCBITaxon:7955` (zebrafish), `NCBITaxon:7227` (Drosophila), `NCBITaxon:6239` (C. elegans). |
-| `assay` | **EFO** | Experimental Factor Ontology, e.g. `EFO:0022605`. |
-| `disease` | **MONDO** | Use `PATO:0000461` for normal/healthy and `MONDO:0021178` for injury. |
-| `development_stage` | organism-specific | **HsapDv** (human), **MmusDv** (mouse), **WBls** (C. elegans), **ZFS** (zebrafish), **FBdv** (Drosophila); `UBERON:0000105` (life cycle stage) for other organisms. Use `unknown` if unavailable and `na` for cell lines. |
-| `tissue` | depends on `type` | **UBERON** for tissue/organoid (or organism-specific **WBbt** / **ZFA** / **FBbt**); **CL** for cell culture; **Cellosaurus** (`CVCL_` prefix) for cell lines; `GO:0005575` (cellular_component) descendants for organelles. |
+| Field | Recommended ontology | Notes & special values                                                                                                                                                                                                       |
+|---|--------------------|------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
+| `organism` | **NCBITaxon**      | e.g. `NCBITaxon:9606` (human), `NCBITaxon:10090` (mouse), `NCBITaxon:7955` (zebrafish), `NCBITaxon:7227` (Drosophila), `NCBITaxon:6239` (C. elegans).                                                                        |
+| `assay` | modality-specific  | Default: Experimental Factor Ontology, e.g. `EFO:0022605`, for imaging use Biological Imaging Methods Ontology eg: `FBbi:00000243`                                                                                                          |
+| `disease` | **MONDO**          | Use `PATO:0000461` for normal/healthy and `MONDO:0021178` for injury.                                                                                                                                                        |
+| `development_stage` | organism-specific  | **HsapDv** (human), **MmusDv** (mouse), **WBls** (C. elegans), **ZFS** (zebrafish), **FBdv** (Drosophila); `UBERON:0000105` (life cycle stage) for other organisms. Use `unknown` if unavailable and `na` for cell lines.    |
+| `tissue` | depends on `type`  | **UBERON** for tissue/organoid (or organism-specific **WBbt** / **ZFA** / **FBbt**); **CL** for cell culture; **Cellosaurus** (`CVCL_` prefix) for cell lines; `GO:0005575` (cellular_component) descendants for organelles. |
 
 The `tissue` entry's `type` field is a controlled value: one of `tissue`, `organoid`,
 `cell culture`, `cell line`, or `organelle`.