Add is_c60_centroid and structure_path to orfs/
Adds two columns to the orfs/ spine, directly after is_c90_centroid:
is_c60_centroid(bool): true for the 1,814,296 orf_ids indims/clusters60.centroid_orf_id.structure_path(string): S3 URI of the centroid's predicted structure, set for 1,813,242 rows;
null elsewhere (including the 1,054 centroids over 2,500 aa, which were not folded).
Structures: s3://petadex-protein-structures/petadex-v1.1-60pid-centroids-structures/ (ESMFold2 Full/Fast, 32 steps, 3 loops, plus AFDB v4 models where the centroid's exact
sequence is in AFDB). Every other column is unchanged: each part was checked for identical row count, orf_id
order and an order-independent hash over all original columns. All 131 parts are rewritten
(307,155,746 rows). The dataset card documents the new columns.
Made by resources/260806_issue6_purav_foldrun/hf_add_structure_columns.py in petadex/igem-toronto
(notebook 260806_issue6_purav_foldrun.ipynb, Part III).
Verified against main: schema, row counts and per-part row order unchanged; is_c60_centroid matches dims/clusters60 exactly; all 1,813,242 structure_paths match structure_index.tsv and sit only on 60% centroids; the 1,054 nulls are all >2,500 aa; the 27 inexact-AFDB centroids point to ESMFold refolds. Added a card commit: orfs/ is not globally sorted by orf_id, and afdb_structures is documented as an accession match. Merging.