annotations: reformat Annotation_Definitions to 'definition ==accession'

Change the annogroup definition format from 'accession=definition' to
'definition ==accession' (literal ' ==' separator: space + two equals signs),
so the column reads as human text with the Pfam accession appended for
provenance. E.g. 'PF00001=7 transmembrane receptor (rhodopsin family)' becomes
'7 transmembrane receptor (rhodopsin family) ==PF00001'. Accessions also still
stand alone in the Annotation_Accessions column.

BLOCK_annotations_X_ocl (the formatter / source of truth):
- utils_run_summary.format_annotation_definitions now emits 'def ==acc'
  (missing definition -> '==acc'); docstrings + the self-documenting column
  headers in scripts 001-004 updated to match.
- Existing RUN_02 outputs (the canonical, server-published run) were backfilled
  in place by a one-off idempotent migration: 945 of 1155 definition-bearing
  TSVs rewritten across all 105 structures (the 210 untouched are the empty
  'zero'-subtype definition columns); full re-scan confirms 0 untransformed
  pairs. A pipeline re-run would reproduce the identical text.

integrator/BLOCK_annotations_X_orthogroups:
- Script 003 no longer strips the accession; it passes the upstream OCL
  definitions through verbatim (so Table 1 inherits 'def ==acc'). RUN_1 re-run,
  validation PASS, re-published.

Also fixes a §39 canonical-RUN violation surfaced while investigating: the OCL
data server was being fed the stale RUN_01 (2026-05-05, pre-definitions) because
RUN_01 kept a live upload_manifest.tsv while the canonical RUN_02's manifest sat
at template defaults (all 'no'). RUN_02's manifest is now switched on (21 yes
rows), the stale RUN_01 manifest removed, and the server re-published from RUN_02
(32,130 files; 16,695 stale RUN_01 symlinks cleaned). The server now serves the
canonical run with definitions in the new format. (RUN manifests + run outputs
are gitignored runtime; only the COPYME source scripts are committed here.)

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Author: 000specific

gigantic_project-COPYME/subprojects/integrator/BLOCK_annotations_X_orthogroups/workflow-COPYME-annotations_X_orthogroups/ai/scripts/003_ai-python-build_annogroup_X_orthogroups.py

@@ -71,27 +71,6 @@ def load_protein_to_orthogroup( orthogroups_path: Path ) -> dict:
     return proteins___orthogroups
 
 
-def strip_accession_prefixes( definitions: str ) -> str:
-    """
-    Drop the leading 'ACCESSION=' from each semicolon-delimited definition pair,
-    keeping only the human-readable definition text (the accessions remain in the
-    separate Annotation_Accessions column).
-
-    'PF00001=7 transmembrane receptor (rhodopsin family); PF13927=Immunoglobulin domain'
-      -> '7 transmembrane receptor (rhodopsin family); Immunoglobulin domain'
-    """
-    if not definitions:
-        return definitions
-    cleaned_definitions = []
-    for pair in definitions.split( ';' ):
-        pair = pair.strip()
-        if not pair:
-            continue
-        # split on the FIRST '=' only — definition text may itself contain '='
-        cleaned_definitions.append( pair.split( '=', 1 )[ 1 ] if '=' in pair else pair )
-    return '; '.join( cleaned_definitions )
-
-
 def load_annogroup_annotations( summary_path: Path ) -> dict:
     """
     annogroup_id -> ( subtype, accessions, definitions ) from the all-types OCL
@@ -182,7 +161,7 @@ def main():
         "Annogroup_ID (annogroup identifier format annogroup_pfam_N)",
         "Annogroup_Subtype (single or combo)",
         "Annotation_Accessions (comma delimited pfam accessions for this annogroup)",
-        "Annotation_Definitions (semicolon delimited pfam definition text for this annogroup; accession prefixes removed, accessions are in Annotation_Accessions)",
+        "Annotation_Definitions (semicolon delimited definition ==accession pairs for this annogroup passed through from the upstream annogroup OCL summary)",
         "Annogroup_Species_Count (number of unique species in the annogroup)",
         "Annogroup_Member_Protein_Count (number of member proteins in the annogroup)",
         "Members_With_Orthogroup_Count (count of member proteins that map to an orthogroup)",
@@ -255,9 +234,9 @@ def main():
                 subtype_from_summary, accessions, definitions = annogroups___annotations.get(
                     annogroup_id, ( subtype, "", "" )
                 )
-                # Definitions carry only the human-readable text — the pfam
-                # accessions live in the Annotation_Accessions column.
-                definitions = strip_accession_prefixes( definitions )
+                # Definitions are passed through verbatim from the upstream
+                # annogroup OCL summary, which formats them as 'definition ==accession'
+                # (semicolon delimited); accessions also stand alone in Annotation_Accessions.
 
                 all_ogs = sorted( orthogroup_ids )
                 output = '\t'.join( [

gigantic_project-COPYME/subprojects/integrator/BLOCK_annotations_X_orthogroups/workflow-RUN_1-annotations_X_orthogroups-run_summary.md

@@ -1,6 +1,6 @@
 # Workflow Run Summary: species70_pfam_X_OrthoHMM
 
-**Status**: ✅ **SUCCESS (completed 2026-06-17 20:05:41)**
+**Status**: ✅ **SUCCESS (completed 2026-06-17 22:49:34)**
 
 **Run label**: `species70_pfam_X_OrthoHMM`
 **Species set**: `species70`

gigantic_project-COPYME/subprojects/ocl_phylogenetic_structures/BLOCK_annotations_X_ocl/workflow-COPYME-ocl_analysis/ai/scripts/001_ai-python-create_annogroups.py

@@ -1,5 +1,5 @@
 # AI: Claude Code | Opus 4.6 | 2026 April 18 | Purpose: Create annotation groups (annogroups) from annotation files and prepare inputs for OCL analysis
-# AI: Claude Code | Opus 4.8 | 2026 June 05 | Purpose: Capture per-accession descriptions and emit Annotation_Definitions (acc=def) alongside Annotation_Accessions on annogroup outputs
+# AI: Claude Code | Opus 4.8 | 2026 June 05 | Purpose: Capture per-accession descriptions and emit Annotation_Definitions (def ==acc) alongside Annotation_Accessions on annogroup outputs
 # Human: Eric Edsinger
 
 """
@@ -926,7 +926,7 @@ def write_annogroup_map( annogroup_map_entries ):
         output += 'Annogroup_Subtype (single or combo or zero)\t'
         output += 'Annotation_Database (name of annotation database)\t'
         output += 'Annotation_Accessions (comma delimited annotation accessions from the database e.g. Pfam PF00069 or unannotated identifier for zero subtype)\t'
-        output += 'Annotation_Definitions (semicolon delimited accession=definition pairs where definition is the InterProScan signature description e.g. PF00069=Protein kinase domain)\t'
+        output += 'Annotation_Definitions (semicolon delimited definition ==accession pairs where definition is the InterProScan signature description e.g. Protein kinase domain ==PF00069)\t'
         output += 'Species_Count (number of unique species with at least one member sequence)\t'
         output += 'Sequence_Count (total number of member sequences)\t'
         output += 'Species_List (comma delimited list of species names as Genus_species)\t'
@@ -968,7 +968,7 @@ def write_annogroups_standardized( annogroup_map_entries ):
         output += 'Species_Count (number of unique species in annogroup)\t'
         output += 'Species_List (comma delimited list of species names as Genus_species)\t'
         output += 'Annotation_Accessions (comma delimited annotation accessions from the database e.g. Pfam PF00069 or unannotated identifier for zero subtype)\t'
-        output += 'Annotation_Definitions (semicolon delimited accession=definition pairs where definition is the InterProScan signature description e.g. PF00069=Protein kinase domain)\n'
+        output += 'Annotation_Definitions (semicolon delimited definition ==accession pairs where definition is the InterProScan signature description e.g. Protein kinase domain ==PF00069)\n'
         output_file.write( output )
 
         for entry in annogroup_map_entries:
@@ -998,7 +998,7 @@ def write_per_subtype_annogroup_files( subtypes___annogroup_data ):
             output += 'Species_List (comma delimited list of species names as Genus_species)\t'
             output += 'Sequence_IDs (comma delimited list of sequence identifiers)\t'
             output += 'Annotation_Accessions (comma delimited annotation accessions from the database e.g. Pfam PF00069 or unannotated identifier for zero subtype)\t'
-            output += 'Annotation_Definitions (semicolon delimited accession=definition pairs where definition is the InterProScan signature description e.g. PF00069=Protein kinase domain)\n'
+            output += 'Annotation_Definitions (semicolon delimited definition ==accession pairs where definition is the InterProScan signature description e.g. Protein kinase domain ==PF00069)\n'
             output_file.write( output )
 
             for entry in annogroup_entries:

gigantic_project-COPYME/subprojects/ocl_phylogenetic_structures/BLOCK_annotations_X_ocl/workflow-COPYME-ocl_analysis/ai/scripts/002_ai-python-determine_origins.py

@@ -777,7 +777,7 @@ def write_annogroup_origins( annogroup_origins, clade_id_names___phylogenetic_bl
             'Species_Count (total unique species in annogroup)',
             'Species_List (comma delimited list of species in annogroup)',
             'Annotation_Accessions (comma delimited annotation accessions from the database e.g. Pfam PF00069 or unannotated identifier for zero subtype)',
-            'Annotation_Definitions (semicolon delimited accession=definition pairs where definition is the InterProScan signature description e.g. PF00069=Protein kinase domain)'
+            'Annotation_Definitions (semicolon delimited definition ==accession pairs where definition is the InterProScan signature description e.g. Protein kinase domain ==PF00069)'
         ]
 
         # Single-row header

gigantic_project-COPYME/subprojects/ocl_phylogenetic_structures/BLOCK_annotations_X_ocl/workflow-COPYME-ocl_analysis/ai/scripts/003_ai-python-quantify_conservation_loss.py

@@ -814,7 +814,7 @@ def write_annogroup_patterns( annogroup_patterns, annogroup_ids___annotation_col
             'Continued_Absence_Events (count of phylogenetic blocks in block-state X where annogroup is absent at both parent and child after an upstream loss)',
             'Species_List (comma delimited list of all species containing this annogroup)',
             'Annotation_Accessions (comma delimited annotation accessions from the database e.g. Pfam PF00069 or unannotated identifier for zero subtype)',
-            'Annotation_Definitions (semicolon delimited accession=definition pairs where definition is the InterProScan signature description e.g. PF00069=Protein kinase domain)'
+            'Annotation_Definitions (semicolon delimited definition ==accession pairs where definition is the InterProScan signature description e.g. Protein kinase domain ==PF00069)'
         ]
 
         # Write single-row header

gigantic_project-COPYME/subprojects/ocl_phylogenetic_structures/BLOCK_annotations_X_ocl/workflow-COPYME-ocl_analysis/ai/scripts/004_ai-python-comprehensive_ocl_analysis.py

@@ -903,7 +903,7 @@ def write_annogroup_summaries( annogroup_summaries, output_file_path, annogroup_
             'Continued_Absence_Events (count of phylogenetic blocks in block-state X where annogroup is absent at both parent and child after an upstream loss)',
             'Species_List (comma delimited list of all species containing this annogroup)',
             'Annotation_Accessions (comma delimited annotation accessions from the database e.g. Pfam PF00069 or unannotated identifier for zero subtype)',
-            'Annotation_Definitions (semicolon delimited accession=definition pairs where definition is the InterProScan signature description e.g. PF00069=Protein kinase domain)'
+            'Annotation_Definitions (semicolon delimited definition ==accession pairs where definition is the InterProScan signature description e.g. Protein kinase domain ==PF00069)'
         ]
 
         csv_writer.writerow( header_columns )

gigantic_project-COPYME/subprojects/ocl_phylogenetic_structures/BLOCK_annotations_X_ocl/workflow-COPYME-ocl_analysis/ai/scripts/utils_run_summary.py

@@ -139,9 +139,9 @@ def clear_fragments_directory( workflow_directory ):
 # Script 001 writes the authoritative annogroup map with two annotation columns:
 #   Annotation_Accessions  -- comma delimited database accessions (e.g. Pfam PF00069)
 #                             or the unannotated identifier for the zero subtype
-#   Annotation_Definitions -- semicolon delimited accession=definition pairs, where
+#   Annotation_Definitions -- semicolon delimited "definition ==accession" pairs, where
 #                             definition is the InterProScan signature description
-#                             (e.g. "PF00069=Protein kinase domain; PF00400=WD40 repeat")
+#                             (e.g. "Protein kinase domain ==PF00069; WD40 repeat ==PF00400")
 #
 # Downstream scripts (002, 003, 004) carry these two columns onto every output
 # that bears an Annogroup_ID by looking them up here by Annogroup_ID. They are
@@ -155,7 +155,7 @@ def sanitize_annotation_text( text ):
 
     Tabs and newlines would break the column/row structure, so they are
     collapsed to spaces. Commas, semicolons and '=' are left intact (the
-    accession=definition; ... format relies on them and TSV only splits on
+    "definition ==accession" ... format relies on them and TSV only splits on
     tabs). Returns the cleaned string.
     """
     if text is None:
@@ -179,9 +179,12 @@ def format_annotation_definitions( accessions, accessions___descriptions ):
         accessions___descriptions: dict mapping accession -> definition text.
 
     Returns:
-        str: semicolon delimited 'accession=definition' pairs over the UNIQUE
-             accessions, e.g. 'PF00069=Protein kinase domain; PF00400=WD40 repeat'.
-             Missing definitions render as 'accession=' (empty definition).
+        str: semicolon delimited 'definition ==accession' pairs over the UNIQUE
+             accessions, e.g. 'Protein kinase domain ==PF00069; WD40 repeat ==PF00400'.
+             Note the literal ' ==' separator (space + two equals signs); the
+             definition comes first so the column reads as human text with the
+             accession appended for provenance. Missing definitions render as
+             '==accession' (no leading text).
     """
     seen_accessions = set()
     pairs = []
@@ -190,7 +193,7 @@ def format_annotation_definitions( accessions, accessions___descriptions ):
             continue
         seen_accessions.add( accession )
         definition = sanitize_annotation_text( accessions___descriptions.get( accession, '' ) )
-        pairs.append( f"{accession}={definition}" )
+        pairs.append( f"{definition} =={accession}" if definition else f"=={accession}" )
     return '; '.join( pairs )