{"success":true,"data":{"author":"theGreatHerrLebert (Claudius timsTOF corpus)","bucketAccessType":"public","bucketName":"theGreatHerrLebert/timstof-dda-pasef-cc0","bucketPrefix":"","category":"Proteomics","cloudProvider":"huggingface","createdAt":"2026-11-17T14:12:21.999Z","description":"A reference corpus of peptide-spectrum matches from timsTOF DDA-PASEF runs, built to train and benchmark peptide property predictors (fragment intensity, ion mobility and collision cross section, retention time, charge) and to configure timsTOF simulators. It covers 58 PRIDE datasets, each verified as CC0 with per-dataset evidence and retrieval dates kept, acquired on timsTOF Pro, Pro 2, HT, SCP, fleX and Ultra instruments, spanning tryptic, HLA immunopeptidomics and PTM-enriched workflows. Tier 1 (tier1/) has 33,717,207 accepted precursor identifications with sequence and modifications, charge, m/z, RT apex and Sage cross-run aligned RT, ion mobility, per-PASEF-event collision energy, RT and ion-mobility peak-shape labels, isotope envelopes, per-engine scores and Sage plus FragPipe agreement. Tier 3 (tier3/) has 504,097,214 matched b/y fragments with calculated and experimental m/z, intensity and per-engine match flags. Inclusion is a union of engines at q <= 0.01, so filter with the provided columns; SCHEMA.md documents fields and manifest.json per-dataset counts. About 33 GB of Parquet with fixed train/validation/test splits. Released under CC0.","displayTitle":"timsTOF DDA-PASEF PSM corpus: 33.7M precursors and 504M fragments from 58 CC0 PRIDE datasets","huggingFaceUrl":"https://huggingface.co/datasets/theGreatHerrLebert/timstof-dda-pasef-cc0","id":"e6c0ebf5-166d-55f4-9332-443390912e11","license":"CC0-1.0","licenseEvidenceUrl":"https://huggingface.co/datasets/theGreatHerrLebert/timstof-dda-pasef-cc0","objectCountEstimate":10,"probeStatus":"ok","region":"us","schemaCard":{"accessHints":[{"language":"python-datasets","snippet":"from datasets import load_dataset\nds = load_dataset(\"theGreatHerrLebert/timstof-dda-pasef-cc0\")"},{"language":"python-hub","snippet":"from huggingface_hub import snapshot_download\nsnapshot_download(repo_id=\"theGreatHerrLebert/timstof-dda-pasef-cc0\", repo_type=\"dataset\")"}],"caveats":["HuggingFace metadata: author=theGreatHerrLebert, license:cc0-1.0, size_categories:100M<n<1B, format:parquet, downloads=131"],"format":"mixed","generatedAt":"2026-11-17T14:23:19.771Z","layout":{"example":"tier1/test.parquet","pattern":"{repoId}/{filename}"},"mountPath":"/mnt/archive","sampleHeaders":[{"bytes":"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","encoding":"hex","key":"tier1/test.parquet","truncatedAt":512}],"sampleObjects":["SCHEMA.md","tier1/test.parquet","tier1/train.parquet","tier1/val.parquet","tier3/test.parquet"],"sidecars":[{"bytes":"---\nlicense: cc0-1.0\npretty_name: Claudius timsTOF DDA-PASEF PSM corpus\ntags:\n  - proteomics\n  - mass-spectrometry\n  - timsTOF\n  - DDA-PASEF\n  - peptide-property-prediction\n  - fragment-intensity\n  - retention-time\n  - ion-mobility\n  - collision-energy\nsize_categories:\n  - 10M<n<100M\nconfigs:\n  - config_name: tier1_psms\n    data_files:\n      - split: train\n        path: tier1/train.parquet\n      - split: validation\n        path: tier1/val.parquet\n      - split: test\n        path: tier1/test.parquet\n  - config_name: tier3_fragments\n    data_files:\n      - split: train\n        path: tier3/train.parquet\n      - split: validation\n        path: tier3/val.parquet\n      - split: test\n        path: tier3/test.parquet\n---\n\n# Claudius timsTOF DDA-PASEF PSM corpus\n\nA large, **CC0**, multi-workflow corpus of peptide-spectrum matches from\n**timsTOF DDA-PASEF** runs, built to train and benchmark peptide-property\npredictors (fragment intensity, ion mobility / CCS, retention time, charge) and\nto configure timsTOF simulators.\n\n> **Design principle — a reference corpus, not a pre-baked training set.**\n> We include broadly, filter minimally at build, and **label richly** so you can\n> reproduce any filtering decision yourself. Every confidence/quality signal is\n> a column; nothing you might want to filter on was thrown away.\n\n## What's novel\n\nTo our knowledge the first public timsTOF DDA-PASEF PSM resource that ships:\n\n1. **Dual-engine (Sage + FragPipe) agreement** labels per PSM and per fragment;\n2. per-precursor **RT and ion-mobility peak shapes** (FWHM, σ, skew, fit R²,\n   trace SNR) — not just apex point estimates;\n3. per-precursor, **per-PASEF-event collision energies** (raw volts);\n4. **matcher-consistent dual-engine b/y fragment** intensities.\n\n## Cohort\n\n**58 datasets**, 100% **CC0** (PRIDE-verified, per-dataset evidence + retrieval\ndates retained), **timsTOF family** instruments (Pro / Pro2 / HT / SCP / fleX /\nUltra), DDA / PASEF acquisition. Tryptic, HLA / immunopeptidomics, and several\nPTM-enriched workflows.\n\n| Tier | rows | train / validation / test |\n|---|---:|---|\n| **Tier 1** — precursors | **33,717,207** | 27,128,597 / 3,205,642 / 3,382,968 |\n| **Tier 3** — b/y fragments | **504,097,214** | 404,005,449 / 48,374,948 / 51,716,817 |\n\n(Per-dataset counts + provenance: `manifest.json`.)\n\n## Tiers\n\n- **`tier1_psms`** — one row per accepted precursor identification\n  `(raw_file, precursor_id)`: sequence + modifications, charge, m/z, RT apex\n  (raw + Sage cross-run-aligned `rt_aligned`), ion mobility, per-PASEF-event\n  collision energy, **RT/IM peak-shape labels** (`ms1_{rt,im}_{fwhm,sigma,skew,r2,snr}`),\n  isotope envelope, per-engine scores, engine-agreement, and derived\n  `width_reliable` / `strict` presets.\n- **`tier3_fragments`** — one row per matched **b/y** fragment (imspy-rematched\n  for both engines under identical settings): calculated + experimental m/z,\n  intensity, ppm error, per-engine match flags, Sage-native provenance.\n\nThe full field-by-field schema (units, null semantics) is in `SCHEMA.md`.\n\n## Filtering & certainty policy (read this before filtering)\n\n- **Inclusion floor:** rank-1 **target** PSMs accepted by **at least one engine\n  at that engine's reported q ≤ 0.01**. ⚠️ This is a *union of two\n  engine-specific 1%-FDR sets* — **not** a guaranteed corpus-level 1% FDR. Treat\n  the per-engine `q`/`pep` as the confidence signals.\n- **Filter up, freely:** every score is exposed (`sage_qvalue`, `sage_pep`,\n  `fragpipe_qvalue`, `fragpipe_pep`, `fragpipe_probability`, hyperscores,\n  `n_engines`, `sage_cosine`, `isotope_cosim`). Applying a stricter reported-q\n  threshold *selects* higher-confidence PSMs — it does **not** recompute FDR;\n  post-hoc combinations don't retain a nominal FDR.\n- **Engine agreement is a label, not a gate** — single-engine PSMs are kept and\n  flagged (`n_engines == 1`). Want the high-confidence subset? Use the\n  documented **`strict`** boolean column.\n- **Peak-shape reliability:** width labels carry `ms1_*_r2` + `ms1_*_snr` + the\n  `*_width_reliable` flags. ⚠️ The ","encoding":"utf8","key":"README.md","name":"README.md","size":6981,"truncated":true},{"bytes":"{\n  \"corpus_version\": \"v0.2\",\n  \"built_unix\": 1781312420,\n  \"split\": {\n    \"method\": \"peptide_hash(sequence_normalized)\",\n    \"seed\": 0,\n    \"fracs\": {\n      \"train\": 0.8,\n      \"val\": 0.1,\n      \"test\": 0.1\n    },\n    \"note\": \"no peptide leakage; group/dataset-level split deferred to a future version\"\n  },\n  \"tier1\": {\n    \"n_datasets\": 58,\n    \"split_rows\": {\n      \"train\": 27180882,\n      \"val\": 3265602,\n      \"test\": 3270723\n    },\n    \"total_rows\": 33717207,\n    \"per_dataset_rows\": {\n      \"PXD019086\": 3848,\n      \"PXD021789\": 2123633,\n      \"PXD026463\": 38347,\n      \"PXD031371\": 1010693,\n      \"PXD031993\": 2434901,\n      \"PXD032159\": 2920352,\n      \"PXD033508\": 1008941,\n      \"PXD035301\": 244666,\n      \"PXD035977\": 282214,\n      \"PXD037501\": 151234,\n      \"PXD037861\": 872924,\n      \"PXD038273\": 119892,\n      \"PXD039787\": 69,\n      \"PXD040716\": 18536,\n      \"PXD040718\": 5609,\n      \"PXD040813\": 969575,\n      \"PXD041154\": 169943,\n      \"PXD042023\": 1377380,\n      \"PXD044326\": 28936,\n      \"PXD046777\": 47323,\n      \"PXD047258\": 20334,\n      \"PXD047581\": 1260266,\n      \"PXD050274\": 1325251,\n      \"PXD050342\": 632318,\n      \"PXD050743\": 1303303,\n      \"PXD052661\": 875622,\n      \"PXD053030\": 10677,\n      \"PXD053346\": 453753,\n      \"PXD053748\": 111026,\n      \"PXD053881\": 2035172,\n      \"PXD054120\": 1024116,\n      \"PXD055279\": 31534,\n      \"PXD055840\": 640004,\n      \"PXD056645\": 70189,\n      \"PXD056651\": 579015,\n      \"PXD058033\": 318582,\n      \"PXD058036\": 118070,\n      \"PXD058623\": 857833,\n      \"PXD058854\": 2052191,\n      \"PXD059051\": 177736,\n      \"PXD060091\": 130507,\n      \"PXD060098\": 31348,\n      \"PXD060682\": 523164,\n      \"PXD061039\": 1269470,\n      \"PXD061057\": 92480,\n      \"PXD064837\": 19961,\n      \"PXD065109\": 90126,\n      \"PXD065448\": 944560,\n      \"PXD065562\": 2060,\n      \"PXD065685\": 168212,\n      \"PXD066365\": 829604,\n      \"PXD068118\": 41244,\n      \"PXD068808\": 233650,\n      \"PXD069251\": 761979,\n      \"PXD069593\": 15375,\n      \"PXD071403\": 435028,\n      \"PXD072060\": 64555,\n      \"PXD072598\": 337876\n    }\n  },\n  \"tier3\": {\n    \"n_datasets\": 58,\n    \"split_rows\": {\n      \"train\": 406006899,\n      \"val\": 49211059,\n      \"test\": 48879256\n    },\n    \"total_rows\": 504097214,\n    \"per_dataset_rows\": {\n      \"PXD019086\": 63424,\n      \"PXD021789\": 37272728,\n      \"PXD026463\": 379394,\n      \"PXD031371\": 14781140,\n      \"PXD031993\": 32626825,\n      \"PXD032159\": 37818188,\n      \"PXD033508\": 6083285,\n      \"PXD035301\": 5525069,\n      \"PXD035977\": 4876186,\n      \"PXD037501\": 1850097,\n      \"PXD037861\": 13940682,\n      \"PXD038273\": 1336743,\n      \"PXD039787\": 1860,\n      \"PXD040716\": 412858,\n      \"PXD040718\": 73341,\n      \"PXD040813\": 15859752,\n      \"PXD041154\": 2862246,\n      \"PXD042023\": 24319319,\n      \"PXD044326\": 374426,\n      \"PXD046777\": 750413,\n      \"PXD047258\": 234315,\n      \"PXD047581\": 19095899,\n      \"PXD050274\": 23201205,\n      \"PXD050342\": 8715017,\n      \"PXD050743\": 21327774,\n      \"PXD052661\": 12482350,\n      \"PXD053030\": 107678,\n      \"PXD053346\": 7398939,\n      \"PXD053748\": 1638696,\n      \"PXD053881\": 34351456,\n      \"PXD054120\": 17574538,\n      \"PXD055279\": 309361,\n      \"PXD055840\": 9927295,\n      \"PXD056645\": 849616,\n      \"PXD056651\": 8308350,\n      \"PXD058033\": 3504688,\n      \"PXD058036\": 1226844,\n      \"PXD058623\": 11510888,\n      \"PXD058854\": 31434908,\n      \"PXD059051\": 2860293,\n      \"PXD060091\": 1618670,\n      \"PXD060098\": 306755,\n      \"PXD060682\": 7044579,\n      \"PXD061039\": 17398816,\n      \"PXD061057\": 1475592,\n      \"PXD064837\": 225509,\n      \"PXD065109\": 1336618,\n      \"PXD065448\": 12799813,\n      \"PXD065562\": 41155,\n      \"PXD065685\": 2859918,\n      \"PXD066365\": 12221927,\n      \"PXD068118\": 372660,\n      \"PXD068808\": 2691749,\n      \"PXD069251\": 14008179,\n      \"PXD069593\": 158894,\n      \"PXD071403\": 6558501,\n      \"PXD072060\": 832483,\n      \"PXD072598\": 4877310\n    }\n  },\n  \"tier1_build_stats\": {\n    \"n_manifests\": 58,\n    \"quarantined\": [],\n    \"bad_manifests\": [],\n    \"n_passed_floor\": 33717207,\n    \"n_written\": 33717207,\n    \"skipped\": {},\n    \"failures\": {}\n  },\n  \"tier3_build_stats","encoding":"utf8","key":"manifest.json","name":"manifest.json","size":4584,"truncated":true}],"sourceObjectKey":"tier1/test.parquet","starterCell":"from datasets import load_dataset\nds = load_dataset(\"theGreatHerrLebert/timstof-dda-pasef-cc0\")\nprint(ds)","starterSamplePath":"tier1/test.parquet","totalObjects":10},"summary":"A reference corpus of peptide-spectrum matches from timsTOF DDA-PASEF runs, built to train and benchmark peptide property predictors (fragment intensity, ion mobility and…","title":"timsTOF DDA-PASEF PSM corpus: 33.7M precursors and 504M fragments from 58 CC0 PRIDE datasets","updatedAt":"2026-11-17T19:57:55.830Z","visibility":"public","websiteUrl":"https://huggingface.co/datasets/theGreatHerrLebert/timstof-dda-pasef-cc0"}}