drop structureId placeholder axis from scores PFrame

The scores PFrame used to declare a `pl7.app/structure/id` axis with a
constant `"static"` value on every row. It was added as a defensive
workaround for what we thought was an xsv-import-pt "at least one axis"
requirement back when the scores frame was emitted via the legacy
single-PDB path. On the PrimaryRef path that's now the only path,
`pframes.processColumn` auto-prepends the upstream PDB column's axes
(`[sampleId, scClonotypeKey]`) to every per-row output, which already
uniquely keys every scores row.

Re-reading `xsv-import-pt.lib.tengo:110-116`, the parquet path's
`PFCONV_IMPORT_CFG_SCHEMA` accepts an empty `axes` array and the
generated polars schema is built from `columns` only when axes is
empty. So the placeholder column was rendering for no reason — it
showed up in the scoresTable as a redundant "Structure" row-axis
cell with the same string `"static"` on every row.

Changes:
- `workflow/src/specs.lib.tengo` — `scoresAxesSpec` is now `[]`,
  `buildScoresXsvSettings` returns `axes: []`. Comments rewritten so
  the next reader doesn't reintroduce the placeholder.
- `software/src/main.py` — drop `structureId` from `_SCORES_AXES` and
  from the `_build_scores_row` dict, drop the `_PLACEHOLDER_STRUCTURE_ID`
  constant.
- `workflow/src/process-pdb.tpl.tengo` — refresh the header comment so
  it documents the actual current behavior (axes = `[sampleId,
  scClonotypeKey]` from prepend, nothing block-side).

NB for local dev: the software image needs rebuilding for the python
change to land at runtime —
`cd software && PL_DOCKER_BUILD=true PL_DOCKER_NO_AUTOPUSH=true pnpm build`.
The workflow-side empty-axes change is forward-compatible with the old
software (pfconv silently ignored undeclared parquet columns), so a
workflow-only deploy still works against a stale image.

Author: xnacly

software/src/main.py

@@ -134,15 +134,14 @@ def chown_to_host(path: Path) -> None:
     ColumnSchema(id="partnerIcode", type="String"),
 ]
 
-# Per-structure scalar PFrame (spec R38). One row per structure. The
-# `structureId` axis is a placeholder until upstream provides a clonotype
-# key via PrimaryRef (R1-R6). R23 summary counts, R38 motif/score scalars,
-# R24-R30 surface metrics, R36 low-conf fractions, R39 threshold flags
-# all live here. Workflow side annotates `*Flag` columns with
-# `pl7.app/isScore: "true"` (R40); raw metrics ship as plain features.
-_SCORES_AXES = [
-    AxisSchema(id="structureId", type="String"),
-]
+# Per-structure scalar PFrame (spec R38). One row per structure. No
+# block-side axes — `pframes.processColumn` prepends the upstream PDB
+# column's `[sampleId, scClonotypeKey]` axes to every emitted PColumn,
+# which is what keys each row. R23 summary counts, R38 motif/score
+# scalars, R24-R30 surface metrics, R36 low-conf fractions, R39
+# threshold flags all live here. Workflow side annotates `*Flag` columns
+# with `pl7.app/isScore: "true"` (R40); raw metrics ship as plain features.
+_SCORES_AXES: list[AxisSchema] = []
 _SCORES_COLUMNS = [
     # Spec R7 mode
     ColumnSchema(id="mode", type="String"),
@@ -183,9 +182,6 @@ def chown_to_host(path: Path) -> None:
     ColumnSchema(id="cdrh3CompactnessFlag", type="String"),
 ]
 
-# Placeholder axis value for the scores PColumn — replaced by the upstream
-# clonotype key once a PrimaryRef-emitting structure-prediction block lands.
-_PLACEHOLDER_STRUCTURE_ID = "static"
 # Spec R39 — sentinel used when a flag isn't applicable to the current mode.
 _FLAG_SENTINEL = "-"
 
@@ -232,10 +228,6 @@ def _build_scores_row(
     missing_canonical = sum(1 for h in cys_hits if h.cysClass == "disulfide_missing")
 
     return {
-        # Placeholder axis until upstream provides a clonotype key via
-        # PrimaryRef (R1-R6). The workflow swaps this column out for the
-        # real scClonotypeKey axis on the PrimaryRef path.
-        "structureId": _PLACEHOLDER_STRUCTURE_ID,
         "mode": mode,
         # R23 summary counts.
         "extraCysCount": extra_cys,

workflow/src/process-pdb.tpl.tengo

@@ -12,11 +12,9 @@
 // The orchestrator (main.tpl.tengo) converts these into PColumns via target
 // outputs of type Resource (liabilitiesJson) and Xsv parquet (motifs/cys/scores).
 // `pframes.processColumn` prepends the input column's axes
-// (`pl7.app/vdj/scClonotypeKey`) to each output PColumn automatically, so the
-// scores frame's Xsv axes are empty here (the python keeps writing the
-// legacy `structureId="static"` axis column — pfconv ignores undeclared
-// parquet columns, matching the spec's PrimaryRef R1-R6 wording that
-// structureId is replaced by the upstream clonotype key).
+// (`[sampleId, scClonotypeKey]`) to each output PColumn automatically, so
+// the scores frame declares no block-side axes (every row is uniquely
+// keyed by the prepended pair).
 
 self := import("@platforma-sdk/workflow-tengo:tpl")
 ll := import("@platforma-sdk/workflow-tengo:ll")

workflow/src/specs.lib.tengo

@@ -333,22 +333,14 @@ cysColumnSpecs := {
 cysColumnIds := ["cysClass", "chainRole", "bondingState", "rsasa", "sidechainRsasa", "sasa", "sidechainSasa", "partnerChainId", "partnerResSeq", "partnerIcode"]
 
 // ──────────────── Scores PFrame ────────────────
-// processColumn's xsv-import-pt requires at least one declared axis even
-// though the upstream `pl7.app/vdj/scClonotypeKey` axis is auto-prepended.
-// We keep `pl7.app/structure/id` as a "static"-valued placeholder so the
-// xsv settings have a non-empty axes list; the final PColumn axes end up
-// as [scClonotypeKey, structureId="static"]. Marked optional so AG-Grid
-// hides it by default.
-scoresAxesSpec := [
-	{
-		name: "pl7.app/structure/id",
-		type: "String",
-		annotations: {
-			"pl7.app/label": "Structure",
-			"pl7.app/table/visibility": "optional"
-		}
-	}
-]
+// The xsv axes list is empty here — processColumn auto-prepends the
+// upstream PDB column's axes (`[sampleId, scClonotypeKey]`) to every
+// output, which already uniquely keys each row. xsv-import-pt + pfconv
+// accept an empty `axes` array on the parquet path (schema is built
+// from `columns` only), so we drop the historical `structureId="static"`
+// placeholder column that otherwise rendered as a redundant "Structure"
+// row-axis cell with the same constant value on every row.
+scoresAxesSpec := []
 
 // Default visibility + orderPriority per spec R51:
 //   Default visible: mode, structuralDevelopabilityRisk, structuralIntegrityRisk,
@@ -625,15 +617,12 @@ buildCysXsvSettings := func(blockId) {
 	}
 }
 
-// Scores frame: parquet carries a `structureId="static"` column the
-// python writer emits for the xsv.importFile "at least one axis"
-// requirement (see `scoresAxesSpec` comment above). Final PColumn axes:
-// [scClonotypeKey, structureId="static"].
+// Scores frame: empty axes; processColumn prepends `[sampleId,
+// scClonotypeKey]` so each per-clonotype row is uniquely keyed without
+// us declaring anything xsv-side.
 buildScoresXsvSettings := func(blockId) {
 	return {
-		axes: [
-			{ column: "structureId", spec: scoresAxesSpec[0] }
-		],
+		axes: [],
 		columns: buildXsvColumns(scoresColumnIds, scoresColumnSpecs, blockId),
 		storageFormat: "Parquet"
 	}