V_epsilon: author did_v1 active-conversion docs for deprecated families

schemas/V_epsilon/conversions/from_did_v1/_index.md

@@ -44,10 +44,27 @@ field-level changes on top.
 | `hartley_reverse_correlation` | NDIcalc-vis-matlab `neuro/hartley_reverse_correlation` | drafted | [hartley_reverse_correlation.md](hartley_reverse_correlation.md) |
 | `hartley_calc` | NDIcalc-vis-matlab `calc/hartley_calc` | drafted | [hartley_calc.md](hartley_calc.md) |
 | `probe_location` | legacy NDI/DID `probe_location` (V_alpha shape) | drafted | [probe_location.md](probe_location.md) |
-| `treatment` | legacy NDI/DID `treatment` (V_alpha shape) | drafted | [treatment.md](treatment.md) |
 | `ontology_image` | legacy NDI/DID `ontologyImage` (V_alpha shape) | drafted | [ontology_image.md](ontology_image.md) |
 | `ontology_label` | legacy NDI/DID `ontologyLabel` (V_alpha shape) | drafted | [ontology_label.md](ontology_label.md) |
 
+## V_epsilon subject_interaction-family conversions (active)
+
+V_epsilon converts the five deprecated families **actively** into the new
+observation / manipulation / annotation classes (not passively into their
+deprecated shapes). These conversions fan out — one did_v1 document can
+mint several V_epsilon documents (a manipulation plus a `time_reference`,
+a `subject` plus a `group_assignment`, etc.). They are implemented in
+DID-matlab's `+did2/+convert/+migrators/`.
+
+| did_v1 source | V_epsilon target(s) | Status | Doc |
+|---|---|---|---|
+| `treatment` | `temperature_` / `procedural_` / `environmental_manipulation` (+ companion `generic_scalar_observation`) | applied-in-tooling | [treatment.md](treatment.md) |
+| `treatment_drug` | `injection` (kind=drug) (+ `utc_reference`) | applied-in-tooling | [treatment_drug.md](treatment_drug.md) |
+| `virus_injection` | `injection` (kind=virus) (+ `utc_reference`) | applied-in-tooling | [virus_injection.md](virus_injection.md) |
+| `treatment_transfer` | `biological_transfer` (+ `utc_reference`) | applied-in-tooling | [treatment_transfer.md](treatment_transfer.md) |
+| `subject_group` | `subject` (is_group) (+ `group_assignment`) | applied-in-tooling | [subject_group.md](subject_group.md) |
+| `stimulus_bath` | `stimulus_bath` v2 (re-rooted under `bath`) | applied-in-tooling | [stimulus_bath.md](stimulus_bath.md) |
+
 ## Notes
 
 - **Not migrated:** `stimloopsplitter_calc` (deprecated per domain owner

schemas/V_epsilon/conversions/from_did_v1/stimulus_bath.md

@@ -0,0 +1,60 @@
+# Conversion: did_v1 → V_epsilon — `stimulus_bath` (re-rooted under `bath`)
+
+## Identity
+
+- **V_epsilon target class:** `stimulus_bath` (v2.0.0), now a concrete
+  subclass of `bath` (← `pharmacological_manipulation` ← `manipulation`
+  ← `subject_interaction`). Moved from `stable/` to `draft/`.
+- **V_epsilon tier:** `draft`
+- **did_v1 source:**
+  `ndi_common/schema_documents/stimulus/stimulus_bath_schema.json`:
+  `location` (structure), `mixture_table` (CSV); superclasses
+  `[base, epochid]`; depends_on `stimulus_element_id` (req).
+- **Status:** `applied-in-tooling`
+  (`DID-matlab/+did2/+convert/+migrators/stimulus_bath.m`)
+
+## Summary
+
+V_epsilon re-roots `stimulus_bath` into the pharmacological-manipulation
+tree. The legacy `epochid` superclass is dropped (timing now lives in
+`time_reference` documents). Fields redistribute across the inherited
+blocks; the migration is 1:1 (no companion) because the legacy class
+carries neither a subject nor a wall-clock time.
+
+## Field mapping
+
+| did_v1 field | V_epsilon field | Transformation |
+|---|---|---|
+| `mixture_table` | `pharmacological_manipulation.mixture` | CSV → records `{chemical, amount:concentration}`; empty ⇒ one blank backfill record |
+| `location` | `bath.location` | `{ontologyNode→node, name}` |
+| — | `bath.kind` | constant `"drug"` default (legacy has no kind; backfill vehicle/wash/tracer) |
+| `stimulus_element_id` | `stimulus_element_id` | carried |
+| (epochid fields) | — | dropped; timing → `time_reference` (backfill) |
+| — | `subject_id` | omitted — legacy carries no subject; backfill |
+
+## Default values for new fields
+
+`bath.kind = "drug"`; `stimulus_bath` block is empty (no own fields).
+`subject_id` and `time_reference` dependencies are omitted (the did2
+validator enforces field/block shape, not depends_on cardinality), and
+flagged for curator backfill.
+
+## Worked example
+
+`stimulus_bath` (location LGN, one-row `mixture_table` for a blocker,
+`stimulus_element_id → E`) → `stimulus_bath` v2 with `mixture` on the
+`pharmacological_manipulation` block, `bath.location = LGN`,
+`bath.kind = "drug"`, and `stimulus_element_id → E`.
+
+## Open questions
+
+- `bath.kind` defaulting to `"drug"` is a heuristic; wash/vehicle baths
+  need curator confirmation.
+- Whether `stimulus_bath` should also gain a `stimulus_presentation`
+  link (vs the bare `stimulus_element_id`) is open in `Bath_Proposal.md`.
+
+## Cross-references
+
+- `Bath_Proposal.md`, `Injection_Proposal.md`
+- Migrator: `DID-matlab/src/did/+did2/+convert/+migrators/stimulus_bath.m`
+- General file-handling rules: [`_files.md`](_files.md)

schemas/V_epsilon/conversions/from_did_v1/subject_group.md

@@ -0,0 +1,58 @@
+# Conversion: did_v1 → V_epsilon — `subject_group` → `subject` + `group_assignment`
+
+## Identity
+
+- **V_epsilon target classes:** `subject` (with `is_group = true`) and,
+  when a member is named, a companion `group_assignment`.
+  `subject_group` moves to `deprecated/`.
+- **V_epsilon tier:** `stable` (`subject`), `draft` (`group_assignment`),
+  `deprecated` (`subject_group`)
+- **did_v1 source:**
+  `ndi_common/schema_documents/subject_group_schema.json`: no fields;
+  depends_on `subject_id` (opt) — an optional member reference.
+- **Status:** `applied-in-tooling`
+  (`DID-matlab/+did2/+convert/+migrators/subject_group.m`)
+
+## Summary
+
+With `subject` carrying `is_group`, the standalone `subject_group`
+identity class is redundant (`Placement_and_Group_Assignment_Proposal.md`
+§2). The migration rewrites the group into a `subject{is_group:true}`,
+**carrying the legacy `base.id` forward so references to the group still
+resolve**, and re-expresses any named member as an event-sourced
+`group_assignment`.
+
+## Field mapping
+
+| did_v1 field | V_epsilon field | Transformation |
+|---|---|---|
+| (identity) | `subject` document | `base.id` carried; `subject.is_group = true`, `is_biological = false`, `local_identifier`/`description` empty |
+| depends_on `subject_id` (member) | companion `group_assignment` | `group_assignment.depends_on`: `subject_id` = the member, `group_id` = the new subject's `base.id` |
+| — | `group_assignment.batch_id` | empty |
+| — | `group_assignment.time_reference_#` | omitted — legacy has no assignment time; backfill |
+
+## Default values for new fields
+
+`subject.is_group = true` (the whole point), `is_biological = false`. The
+`subject` keeps the legacy `base.id` and `session_id`; the
+`group_assignment` mints a fresh id and shares the `session_id`.
+
+## Worked example
+
+`subject_group` (id `G`, member `M`) → `subject{is_group:true}` with
+`base.id = G`, plus `group_assignment{subject_id:M, group_id:G}`.
+
+## Open questions
+
+- The real legacy `subject_group` carries at most one optional member; a
+  group with many members was represented by many `subject_group` docs or
+  by member-side links. Multi-member groups may need a corpus-specific
+  pass to mint one `group_assignment` per member.
+- "Migrate-then-tighten": `group_assignment.group_id` targets `subject`;
+  the migration must complete before any tightening of that reference.
+
+## Cross-references
+
+- `Placement_and_Group_Assignment_Proposal.md` §2, `Subject_Document_Proposal.md`
+- Migrator: `DID-matlab/src/did/+did2/+convert/+migrators/subject_group.m`
+- General file-handling rules: [`_files.md`](_files.md)