{"success":true,"data":{"author":"Briney Lab, Scripps Research; structures from the wwPDB","bucketAccessType":"public","bucketName":"brineylab/oplmfold-training-data","bucketPrefix":"","category":"Structural Biology","cloudProvider":"huggingface","createdAt":"2026-11-22T01:09:20.869Z","description":"Structure training data for OPLMFold, the structure-prediction head of the open protein language model OPLM from the Briney Lab at Scripps Research. The first source folder, wwpdb/2026-10-12, holds the protein chains of the wwPDB mmCIF archive as of 12 October 2026: 254,424 entries and 1,080,220 chains (37 GiB) preprocessed into OPLM's flat all-atom representation, with the Chemical Component Dictionary table. Each versioned folder is self-describing, with a README, a manifest.json of provenance, counts and rejected entries, an index.parquet of one row per chain with its split, Parquet shards of one row per entry, held-out target lists and checksums. Splits follow a date cutoff: train is released on or before 2021-09-30, with valid, post_cutoff, CAMEO and CASP14 held-out evaluation targets and pre-cutoff entries that share a sequence with a held-out entry set apart. Use it to train or evaluate protein structure prediction and folding models; load with oplm.fold.data.StructureDataset or read the Parquet directly. wwPDB data are CC0 1.0; cite the wwPDB consortium, Nucleic Acids Res. 47, D520 (2019).","displayTitle":"OPLMFold Training Data: wwPDB Protein Chains (2026-10-12 Snapshot) Preprocessed for Protein Structure Prediction (Briney Lab)","huggingFaceUrl":"https://huggingface.co/datasets/brineylab/oplmfold-training-data","id":"04734bdd-ce00-57df-b83d-f62754a608b8","license":"CC0-1.0","licenseEvidenceUrl":"https://www.wwpdb.org/about/usage-policies","objectCountEstimate":108,"probeStatus":"ok","referenceUrl":"https://github.com/briney/oplm","region":"us","schemaCard":{"accessHints":[{"language":"python-datasets","snippet":"from datasets import load_dataset\nds = load_dataset(\"brineylab/oplmfold-training-data\")"},{"language":"python-hub","snippet":"from huggingface_hub import snapshot_download\nsnapshot_download(repo_id=\"brineylab/oplmfold-training-data\", repo_type=\"dataset\")"}],"caveats":["HuggingFace metadata: author=brineylab, license:cc0-1.0, downloads=37"],"format":"mixed","generatedAt":"2026-11-22T01:10:21.668Z","layout":{"example":"wwpdb/2026-10-12/index.parquet","pattern":"{repoId}/{filename}"},"mountPath":"/mnt/archive","sampleHeaders":[{"bytes":"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","encoding":"hex","key":"wwpdb/2026-10-12/index.parquet","truncatedAt":512}],"sampleObjects":["wwpdb/2026-10-12/ccd_table.json","wwpdb/2026-10-12/checksums.txt","wwpdb/2026-10-12/index.parquet","wwpdb/2026-10-12/shards/shard-00000.parquet","wwpdb/2026-10-12/shards/shard-00001.parquet"],"sidecars":[{"bytes":"---\nlicense: cc0-1.0\nviewer: false\npretty_name: OPLMFold training data\ntags:\n- protein-structure\n- wwpdb\n- biology\n---\n\n# OPLMFold training data\n\nStructural training data for OPLMFold, the structure-prediction head of the\n[open protein language model (OPLM)](https://github.com/briney/oplm). Each source\nof structures is preprocessed once into OPLM's flat all-atom representation\n(`oplm.fold.data.complex.MolecularComplex`) and published here as a self-contained\nfolder that `oplm.fold.data.dataset.StructureDataset` loads directly.\n\n## Inventory\n\n| folder | source | snapshot | entries | chains | size | license | tag |\n|---|---|---|---|---|---|---|---|\n| `wwpdb/2026-10-12/` | wwPDB mmCIF archive, protein chains | 2026-10-12 | 254,424 | 1,080,220 | 37 GiB | CC0 1.0 | `wwpdb-2026-10-12` |\n\nPlanned sources, each in its own folder: `afdb-distill/` (monomer distillation),\n`afdb-dimers/` (the AFDB quaternary-structure release), `antibody-antigen/`.\n\n## Layout rules\n\n- **One folder per source and version**, `<source>/<version>/`. A version folder never\n  changes after its tag is created; a refreshed source is a new sibling folder, so a\n  training config that names `wwpdb/2026-10-12` always means the same bytes.\n- **Every version folder is self-describing**: `README.md`, `manifest.json` (provenance,\n  configuration, counts, rejections), `index.parquet` (one row per chain, with the\n  `split` column), `shards/*.parquet` (one row per entry), `targets/` (held-out target\n  lists), `checksums.txt`.\n- **One tag per publish**, `<source>-<version>`. A tag pins the whole repository as it\n  was at that publish.\n- The frontmatter license is the most permissive statement that covers every folder; the\n  inventory table gives each source's own license.\n\n## Download\n\n```bash\n# everything\nhf download brineylab/oplmfold-training-data --repo-type dataset --local-dir oplmfold-data\n# one source\nhf download brineylab/oplmfold-training-data --repo-type dataset --include \"wwpdb/2026-10-12/*\" --local-dir oplmfold-data\n# only the index and manifest, to inspect before fetching the shards\nhf download brineylab/oplmfold-training-data --repo-type dataset \\\n    --include \"wwpdb/2026-10-12/index.parquet\" --include \"wwpdb/2026-10-12/manifest.json\" --local-dir oplmfold-data\n```\n\n## Load\n\n```bash\npip install \"oplm[fold] @ git+https://github.com/briney/oplm\"\n```\n\n```python\nfrom functools import partial\nfrom torch.utils.data import DataLoader\nfrom oplm.fold.data.collate import collate_rows\nfrom oplm.fold.data.dataset import StructureDataset\n\nds = StructureDataset(\"oplmfold-data/wwpdb/2026-10-12\", split=\"train\", crop_size=384, seed=0)\nloader = DataLoader(ds, batch_size=8, collate_fn=partial(collate_rows, crop_size=384))\n```\n\nSplits in the wwPDB index: `train` (released on or before 2021-09-30), `valid` (64 seeded\npost-cutoff monomers), `post_cutoff`, `cameo` and `casp14` (held-out evaluation targets),\n`excluded_heldout_sequence` (pre-cutoff entries sharing a chain sequence with a held-out\nentry), `undated`. See `wwpdb/<version>/README.md` and the OPLM docs (`docs/FOLD.md` §8)\nfor the parsing rules and the column schema.\n\n## Sources and licenses\n\n- wwPDB archive and the Chemical Component Dictionary: [CC0 1.0](https://www.wwpdb.org/about/usage-policies).\n  wwPDB consortium, *Nucleic Acids Res.* 47, D520 (2019).\n- Held-out target lists: CAMEO (Haas *et al.*, *Proteins* 2018; weekly target exports) and\n  CASP14 (Prediction Center target list and sequences); provenance in each\n  `targets/` folder and in `docs/fold/m2a/` of the OPLM repository.\n","encoding":"utf8","key":"README.md","name":"README.md","size":3555,"truncated":false},{"bytes":"---\nlicense: cc0-1.0\npretty_name: \"OPLM fold data: pdb 2026-10-12\"\ntags:\n- protein-structure\n- wwpdb\n---\n\n# OPLM fold data: pdb 2026-10-12\n\nProtein chains of the wwPDB archive parsed into OPLM's flat all-atom representation\n(`oplm.fold.data.complex.MolecularComplex`) for training the OPLM structure prediction head.\nSnapshot of 2026-10-12; 254424 entries accepted, 1080220 chains; CCD components.cif.gz (51389 components).\nSource data: wwPDB (CC0 1.0) and the wwPDB Chemical Component Dictionary (CC0 1.0).\n\nLayout: `shards/*.parquet` (one row per entry; see `manifest.json` for the schema version and\n`oplm.fold.data.complex.COMPLEX_SCHEMA` for the columns), `index.parquet` (one row per chain\nwith the `split` column), `targets/*.json` (held-out target lists), `checksums.txt`\n(sha256 of every archive entry file; the CCD's sha256 is in `manifest.json` under\n`ccd.sha256`), `manifest.json` (provenance, configuration, counts).\n\nSplits: cameo 744, casp14 128, excluded_heldout_sequence 1239, post_cutoff 75763, train 176486, valid 64\n\nLoad with `oplm.fold.data.dataset.StructureDataset(path, split=\"train\", crop_size=384)`;\nsee `docs/FOLD.md` in https://github.com/briney/oplm.\n","encoding":"utf8","key":"wwpdb/2026-10-12/README.md","name":"README.md","size":1180,"truncated":false},{"bytes":"{\n \"schema_version\": 1,\n \"parse_version\": 1,\n \"source\": \"pdb\",\n \"created\": \"2026-10-12T17:02:11+00:00\",\n \"snapshot_date\": \"2026-10-12\",\n \"archive\": \"mmCIF\",\n \"ccd\": {\n  \"path\": \"components.cif.gz\",\n  \"sha256\": \"405da7e5d06527d45292db9400378d9788c024830cad4e3b43890b6c6a0dd4c3\",\n  \"components\": 51389\n },\n \"inputs\": 260626,\n \"accepted\": 254424,\n \"rejected\": {\n  \"no_protein_chains\": 5265,\n  \"resolution\": 928,\n  \"no_resolved_chain\": 9\n },\n \"rejections\": [\n  {\n   \"id\": \"100d\",\n   \"reason\": \"no_protein_chains\",\n   \"detail\": \"\"\n  },\n  {\n   \"id\": \"101d\",\n   \"reason\": \"no_protein_chains\",\n   \"detail\": \"\"\n  },\n  {\n   \"id\": \"102d\",\n   \"reason\": \"no_protein_chains\",\n   \"detail\": \"\"\n  },\n  {\n   \"id\": \"103d\",\n   \"reason\": \"no_protein_chains\",\n   \"detail\": \"\"\n  },\n  {\n   \"id\": \"104d\",\n   \"reason\": \"no_protein_chains\",\n   \"detail\": \"\"\n  },\n  {\n   \"id\": \"105d\",\n   \"reason\": \"no_protein_chains\",\n   \"detail\": \"\"\n  },\n  {\n   \"id\": \"106d\",\n   \"reason\": \"no_protein_chains\",\n   \"detail\": \"\"\n  },\n  {\n   \"id\": \"107d\",\n   \"reason\": \"no_protein_chains\",\n   \"detail\": \"\"\n  },\n  {\n   \"id\": \"108d\",\n   \"reason\": \"no_protein_chains\",\n   \"detail\": \"\"\n  },\n  {\n   \"id\": \"109d\",\n   \"reason\": \"no_protein_chains\",\n   \"detail\": \"\"\n  },\n  {\n   \"id\": \"10ch\",\n   \"reason\": \"no_protein_chains\",\n   \"detail\": \"\"\n  },\n  {\n   \"id\": \"10fy\",\n   \"reason\": \"no_protein_chains\",\n   \"detail\": \"\"\n  },\n  {\n   \"id\": \"10hc\",\n   \"reason\": \"no_protein_chains\",\n   \"detail\": \"\"\n  },\n  {\n   \"id\": \"10hd\",\n   \"reason\": \"no_protein_chains\",\n   \"detail\": \"\"\n  },\n  {\n   \"id\": \"10he\",\n   \"reason\": \"no_protein_chains\",\n   \"detail\": \"\"\n  },\n  {\n   \"id\": \"10hf\",\n   \"reason\": \"no_protein_chains\",\n   \"detail\": \"\"\n  },\n  {\n   \"id\": \"10hs\",\n   \"reason\": \"no_protein_chains\",\n   \"detail\": \"\"\n  },\n  {\n   \"id\": \"10ht\",\n   \"reason\": \"no_protein_chains\",\n   \"detail\": \"\"\n  },\n  {\n   \"id\": \"10id\",\n   \"reason\": \"resolution\",\n   \"detail\": \"9.54 > 9.0\"\n  },\n  {\n   \"id\": \"10zt\",\n   \"reason\": \"no_protein_chains\",\n   \"detail\": \"\"\n  },\n  {\n   \"id\": \"10zu\",\n   \"reason\": \"no_protein_chains\",\n   \"detail\": \"\"\n  },\n  {\n   \"id\": \"110d\",\n   \"reason\": \"no_protein_chains\",\n   \"detail\": \"\"\n  },\n  {\n   \"id\": \"111d\",\n   \"reason\": \"no_protein_chains\",\n   \"detail\": \"\"\n  },\n  {\n   \"id\": \"112d\",\n   \"reason\": \"no_protein_chains\",\n   \"detail\": \"\"\n  },\n  {\n   \"id\": \"113d\",\n   \"reason\": \"no_protein_chains\",\n   \"detail\": \"\"\n  },\n  {\n   \"id\": \"114d\",\n   \"reason\": \"no_protein_chains\",\n   \"detail\": \"\"\n  },\n  {\n   \"id\": \"115d\",\n   \"reason\": \"no_protein_chains\",\n   \"detail\": \"\"\n  },\n  {\n   \"id\": \"116d\",\n   \"reason\": \"no_protein_chains\",\n   \"detail\": \"\"\n  },\n  {\n   \"id\": \"117d\",\n   \"reason\": \"no_protein_chains\",\n   \"detail\": \"\"\n  },\n  {\n   \"id\": \"118d\",\n   \"reason\": \"no_protein_chains\",\n   \"detail\": \"\"\n  },\n  {\n   \"id\": \"119d\",\n   \"reason\": \"no_protein_chains\",\n   \"detail\": \"\"\n  },\n  {\n   \"id\": \"11ag\",\n   \"reason\": \"no_protein_chains\",\n   \"detail\": \"\"\n  },\n  {\n   \"id\": \"11bj\",\n   \"reason\": \"no_protein_chains\",\n   \"detail\": \"\"\n  },\n  {\n   \"id\": \"11bk\",\n   \"reason\": \"no_protein_chains\",\n   \"detail\": \"\"\n  },\n  {\n   \"id\": \"11eh\",\n   \"reason\": \"no_protein_chains\",\n   \"detail\": \"\"\n  },\n  {\n   \"id\": \"11xu\",\n   \"reason\": \"resolution\",\n   \"detail\": \"9.6 > 9.0\"\n  },\n  {\n   \"id\": \"121d\",\n   \"reason\": \"no_protein_chains\",\n   \"detail\": \"\"\n  },\n  {\n   \"id\": \"122d\",\n   \"reason\": \"no_protein_chains\",\n   \"detail\": \"\"\n  },\n  {\n   \"id\": \"123d\",\n   \"reason\": \"no_protein_chains\",\n   \"detail\": \"\"\n  },\n  {\n   \"id\": \"124d\",\n   \"reason\": \"no_protein_chains\",\n   \"detail\": \"\"\n  },\n  {\n   \"id\": \"126d\",\n   \"reason\": \"no_protein_chains\",\n   \"detail\": \"\"\n  },\n  {\n   \"id\": \"127d\",\n   \"reason\": \"no_protein_chains\",\n   \"detail\": \"\"\n  },\n  {\n   \"id\": \"128d\",\n   \"reason\": \"no_protein_chains\",\n   \"detail\": \"\"\n  },\n  {\n   \"id\": \"129d\",\n   \"reason\": \"no_protein_chains\",\n   \"detail\": \"\"\n  },\n  {\n   \"id\": \"12aa\",\n   \"reason\": \"no_protein_chains\",\n   \"detail\": \"\"\n  },\n  {\n   \"id\": \"12ae\",\n   \"reason\": \"no_protein_chains\",\n   \"detail\": \"\"\n  },\n  {\n   \"id\": \"12ci\",\n   \"reason\": \"no_protein_chains\",\n   \"detail\": \"\"\n  },\n  {\n   \"id\": \"12cj\",\n   \"reason\": \"no_protein_chains\",\n   \"detai","encoding":"utf8","key":"wwpdb/2026-10-12/manifest.json","name":"manifest.json","size":482412,"truncated":true},{"bytes":"# Held-out target list sources (fetched 2026-10-12)\n\n| file | origin |\n|---|---|\n| cameo-1year-to-2022-01-15.json | http://web.archive.org/web/20230807161458id_/https://cameo3d.org/modeling/targets/1-year/ajax/?to_date=2022-01-15 |\n| cameo-1year-to-2022-12-31.json | https://github.com/Jianxinnn/Biodataset4AI/blob/main/Protein/cameo/2022/raw/cameo_2022_targets_1year_to_2022-12-31.json (their copy of https://archive.cameo3d.org/modeling/targets/1-year/ajax/?to_date=2022-12-31, generated 2026-04-12T12:05:26Z) |\n| cameo-wayback-cdx-ajax.json | Wayback CDX listing showing the 2022-01-15 capture is the only archived 1-year JSON |\n| crosscheck-eigenfold-cameo2022.csv | https://github.com/bjing2016/EigenFold/blob/master/splits/cameo2022.csv (183 CAMEO targets Aug-Oct 2022; all 183 present in source 2) |\n| crosscheck-manyfold-sequences_cameo.fasta | https://github.com/instadeepai/manyfold/blob/main/datasets/sequences_cameo.fasta (143 CAMEO targets Mar-May 2022; all 143 present in source 2) |\n| casp14.seq.txt | https://predictioncenter.org/download_area/CASP14/sequences/casp14.seq.txt |\n| casp14-targetlist.cgi.html | https://predictioncenter.org/casp14/targetlist.cgi?view_targets=all&view=all |\n| casp14-domains_summary.cgi.html | https://predictioncenter.org/casp14/domains_summary.cgi |\n| casp14-targetlist-parsed.json, casp14-official-merged.json | parsed from the two pages above |\n| casp14-rcsb-seqsearch-*.json | https://search.rcsb.org/rcsbsearch/v2/query, service \"sequence\", evalue 0.1, identity 1.0 / 0.95 (verbose) |\n\narchive.cameo3d.org (the CAMEO archive) answered HTTP 503 for every URL on 2026-10-12; www.cameo3d.org\nredirects there. Checksums: SHA256SUMS.\n","encoding":"utf8","key":"wwpdb/2026-10-12/targets/sources/README.md","name":"README.md","size":1681,"truncated":false}],"sourceObjectKey":"wwpdb/2026-10-12/index.parquet","starterCell":"from datasets import load_dataset\nds = load_dataset(\"brineylab/oplmfold-training-data\")\nprint(ds)","starterSamplePath":"wwpdb/2026-10-12/index.parquet","totalObjects":108},"summary":"Structure training data for OPLMFold, the structure-prediction head of the open protein language model OPLM from the Briney Lab at Scripps Research.","title":"OPLMFold Training Data: wwPDB Protein Chains (2026-10-12 Snapshot) Preprocessed for Protein Structure Prediction (Briney Lab)","updatedAt":"2026-11-22T02:33:20.686Z","visibility":"public","websiteUrl":"https://huggingface.co/datasets/brineylab/oplmfold-training-data"}}