{"success":true,"data":{"author":"Corinna Brungs, Robin Schmid, Tomas Pluskal et al. (IOCB Prague)","bucketAccessType":"public","bucketName":"structure-epflai/msnlib","bucketPrefix":"","category":"Chemistry & Materials","cloudProvider":"huggingface","createdAt":"2026-11-22T01:09:20.592Z","description":"MSnLib is an open library of multi-stage fragmentation (MSn) tandem mass spectra for small molecules, built by the Pluskal lab at IOCB Prague by flow-injection analysis of commercial compound libraries on an Orbitrap ID-X and automatic library generation in mzmine. Each compound ion carries MS2 spectra, pseudo-MS2 spectra (a full MSn tree merged into one spectrum) and the individual MS3+ fragmentation stages, in positive and negative ionization modes. This Hugging Face copy holds the 4,532,018 MGF records of Zenodo record 21105617 as Parquet with the original peak order and intensities, precursor m/z, adduct, collision energy and source metadata kept as text, plus a compounds table of the nine detected libraries; MS2 and MSn exports overlap by design, so merged spectra are not independent measurements. Use it to train and benchmark spectrum-to-structure models, spectral similarity search and fragmentation-tree prediction for metabolomics and natural products; load with pandas, pyarrow or datasets. License CC BY 4.0 (Zenodo, DOI 10.5281/zenodo.21105617). Cite Brungs, Schmid, Heuckeroth et al., MSnLib: efficient generation of open multi-stage fragmentation mass spectral libraries, Nature Methods 22, 2028 (2025).","displayTitle":"MSnLib: Multi-Stage (MSn) Tandem Mass Spectral Libraries of 4.5 Million Orbitrap Spectra (IOCB Prague, Parquet)","doi":"10.5281/zenodo.21105617","huggingFaceUrl":"https://huggingface.co/datasets/structure-epflai/msnlib","id":"ba03403f-84b8-5a16-877a-aa8dbddd8e9f","license":"CC-BY-4.0","licenseEvidenceUrl":"https://zenodo.org/records/21105617","objectCountEstimate":125,"probeStatus":"ok","referenceUrl":"https://doi.org/10.1038/s41592-025-02813-0","region":"us","schemaCard":{"accessHints":[{"language":"python-datasets","snippet":"from datasets import load_dataset\nds = load_dataset(\"structure-epflai/msnlib\")"},{"language":"python-hub","snippet":"from huggingface_hub import snapshot_download\nsnapshot_download(repo_id=\"structure-epflai/msnlib\", repo_type=\"dataset\")"}],"caveats":["HuggingFace metadata: author=structure-epflai, license:cc-by-4.0, size_categories:1M<n<10M, format:parquet, downloads=731"],"format":"mixed","generatedAt":"2026-11-22T01:10:19.775Z","layout":{"example":"data/compounds/20250828_9libraries_only_detected_cleaned.parquet","pattern":"{repoId}/{filename}"},"mountPath":"/mnt/archive","sampleHeaders":[{"bytes":"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","encoding":"hex","key":"data/compounds/20250828_9libraries_only_detected_cleaned.parquet","truncatedAt":512}],"sampleObjects":["collate_ms_libraries.py","data/compounds/20250828_9libraries_only_detected_cleaned.parquet","data/msnlib/train-00000.parquet","data/msnlib/train-00001.parquet","data/msnlib/train-00002.parquet"],"sidecars":[{"bytes":"---\nlicense: cc-by-4.0\npretty_name: msnlib\nconfigs:\n- config_name: full\n  default: true\n  data_files:\n  - split: train\n    path: data/msnlib/*.parquet\n- config_name: compounds\n  data_files:\n  - split: train\n    path: data/compounds/*.parquet\n---\n\n# msnlib\n\nPinned source: https://zenodo.org/records/21105617 . This mirror contains 4,532,018 source MGF records with original peak order and intensities, plus original source files. No molecule standardization, chemistry filter, peak normalization, or new benchmark split is applied. `train` is a storage convention.\n\nMetadata values and duplicate metadata keys remain in `source_metadata_json`; source identifiers and record indices are preserved. Raw charge, precursor, adduct, and collision-energy text are retained rather than imputed. Extra peak fields are preserved. Empty/negative-intensity source spectra are flagged and retained.\n\nMS2 and MSn exports overlap by design: source MS2 files include pseudo-MS2 spectra, and MSn files also contain individual higher fragmentation stages. Do not count merged spectra or duplicate export views as independent measurements. `SPECTYPE` is retained and pseudo-MS2 flagged. The `source_file` suffix distinguishes MS2-only/MSn views and polarity. Source compound metadata and both JSON/MGF representations remain under `source/`. The `compounds` configuration exposes the source detected-compound Parquet unchanged.\n\nPublished file MD5 checksums, row counts, peak-vector shapes, and finite values were checked. Full chemical annotation validity and overlap with other datasets remain unaudited. Cite the original source authors and publication described in `source/source_metadata.json`.\n","encoding":"utf8","key":"README.md","name":"README.md","size":1682,"truncated":false},{"bytes":"{\n  \"dataset\": \"msnlib\",\n  \"source_metadata\": {\n    \"record_id\": \"21105617\",\n    \"metadata\": {\n      \"title\": \"MSnLib Mass spectral libraries (.mgf and .json)\",\n      \"doi\": \"10.5281/zenodo.21105617\",\n      \"publication_date\": \"2026-07-01\",\n      \"description\": \"<p>If you use the data of MSnLib, please cite the <a href=\\\"https://doi.org/10.1038/s41592-025-02813-0\\\">MSnLib </a>paper in any resulting publication (Brungs, C., Schmid, R., Heuckeroth, S. et al. MSnLib: efficient generation of open multi-stage fragmentation mass spectral libraries. Nat Methods 22, 2028&ndash;2031 (2025). https://doi.org/10.1038/s41592-025-02813-0)</p>\\n<p>The data for&nbsp;<a href=\\\"https://doi.org/10.1038/s41592-025-02813-0\\\">MSnLib</a> are divided into several Zenodo records due to size constraints.&nbsp;</p>\\n<p>raw positive: <a href=\\\"https://doi.org/10.5281/zenodo.10966404\\\">10966404</a><br>raw negative: <a href=\\\"https://doi.org/10.5281/zenodo.10967081\\\">10967081</a><br>mzml positive and negative: <a href=\\\"https://doi.org/10.5281/zenodo.10966280\\\">10966280</a><br>spectral libraries: <a href=\\\"https://doi.org/10.5281/zenodo.11163380\\\">11163380</a></p>\\n<p>This record includes the automatically generated spectral libraries (MSnLib) within mzmine, acquired using a flow injection method on an Orbitrap ID-X instrument, for all compound libraries. There are multiple files for each compound library containing MS2 only or MSn in two data formats (.mgf or .json) for both polarities. Information about detected compounds are included in the .parquet file.&nbsp;</p>\\n<p>MS2 contains next to all MS2 spectra all pseudo MS2 spectra (a full MSn tree merged into one spectrum per compound ion). MSn contains all individual MSn stages additionally. The first number for each file highlights the library building date.</p>\\n<p>9 Compound Libraries:</p>\\n<ul>\\n<li>Short Name: Full name, Provider (Catalog number), total compounds (not all detected during library building)</li>\\n<li>MCEBIO: Bioactive Compound Library, MedChemExpress (HY-L001), 10,315 compounds</li>\\n<li>MCESAF: 5k Scaffold Library, MedChemExpress, (HY-L902), 4998 compounds</li>\\n<li>NIHNP: NIH NPAC ACONN collection of NP, NIH/NCATS, 3988 compounds</li>\\n<li>OTAVAPEP: Alpha-helix Peptiomimetic Library, OTAVAchemicals (a-helix-Peptido), 1298 compounds</li>\\n<li>ENAMDISC: Discovery Diversity Set -10, Enamine (DDS-10), 10,240 compounds</li>\\n<li>ENAMMOL: Carboxylic Acid Fragment Library + Random, Enamine and Molport, 4378 compounds</li>\\n<li>MCEDRUG: FDA-Approved Drug Library, MedChemExpress (HY-L022), 2610 compounds</li>\\n<li>MCEDIV_50k_Sub: Subset of 50K Diversity Library, MedChemExpress (HY-L901), 20000 compounds</li>\\n<li>TargetMolHTSNP: Subset of Natural Product Library for HTS, TargetMol (L6000), 2175 compounds</li>\\n</ul>\\n<p>Information regarding the SPECTYPE</p>\\n<ul>\\n<li>no SPECTYPE or SINGLE_BEST_SCAN: Best spectrum for each precursor and energy (highest TIC)</li>\\n<li>'SAME_ENERGY' = Additionally, if a spectrum was acquired multiple times for a precursor with the same energy, they are merged into one spectrum only with the same energy (max. signal height used for each fragment signal).</li>\\n<li>'ALL_ENERGIES' = merged spectrum of all used energies (in our case 3 for each precursor, using the merged (same energy) if available).</li>\\n<li>'ALL_MSN_TO_PSEUDO_MS2' = mzmine merges all MSn into one pseudo MS2.</li>\\n</ul>\\n<p>&nbsp;</p>\\n<p>MCEDIV and TargetmolNPHTS&nbsp;</p>\",\n      \"access_right\": \"open\",\n      \"creators\": [\n        {\n          \"name\": \"Brungs, Corinna\",\n          \"affiliation\": \"Czech Academy of Sciences, Institute of Organic Chemistry and Biochemistry\",\n          \"orcid\": \"0000-0002-2571-5235\"\n        },\n        {\n          \"name\": \"Schmid, Robin\",\n          \"affiliation\": \"Czech Academy of Sciences, Institute of Organic Chemistry and Biochemistry\"\n        },\n        {\n          \"name\": \"Pluskal, Tomas\",\n          \"affiliation\": \"Czech Academy of Sciences, Institute of Organic Chemistry and Biochemistry\"\n        }\n      ],\n      \"keywords\": [\n        \"MSnLib\",\n        \"s","encoding":"utf8","key":"manifest.json","name":"manifest.json","size":59119,"truncated":true}],"sourceObjectKey":"data/compounds/20250828_9libraries_only_detected_cleaned.parquet","starterCell":"from datasets import load_dataset\nds = load_dataset(\"structure-epflai/msnlib\")\nprint(ds)","starterSamplePath":"data/compounds/20250828_9libraries_only_detected_cleaned.parquet","totalObjects":125},"summary":"MSnLib is an open library of multi-stage fragmentation (MSn) tandem mass spectra for small molecules, built by the Pluskal lab at IOCB Prague by flow-injection analysis of…","title":"MSnLib: Multi-Stage (MSn) Tandem Mass Spectral Libraries of 4.5 Million Orbitrap Spectra (IOCB Prague, Parquet)","updatedAt":"2026-11-22T04:01:56.617Z","visibility":"public","websiteUrl":"https://huggingface.co/datasets/structure-epflai/msnlib"}}