{"success":true,"data":{"author":"Chandar Lab (Fournier et al.)","bucketAccessType":"public","bucketName":"chandar-lab/UR100P","bucketPrefix":"","category":"Proteomics","cloudProvider":"huggingface","createdAt":"2026-11-17T14:12:21.334Z","description":"The curated protein sequence corpus used to train and evaluate AMPLIFY, the efficient protein language model from Chandar Lab. It combines UniProt (UniRef100) sequences, paired heavy and light antibody chains from the Observed Antibody Space (OAS), and domains from the Structural Classification of Proteins version 2 (SCOP 2), all collected in December 2023. Sequences with ambiguous residues (B, J, O, U, X, Z) were removed. OAS pairs appear in both heavy|light and light|heavy order separated by a chain-break token, and MMseqs2 was used to drop training sequences above 90% identity to the validation sets so that perplexity and downstream evaluations are not inflated by leakage. The repository (about 186 GB) has one folder per source, UniProt/, OAS/ and SCOP/, each with train and test splits; load a subset with Hugging Face datasets, for example load_dataset(\"chandar-lab/UR100P\", data_dir=\"UniProt\", split=\"test\"). Useful for protein language model pretraining, antibody language modeling and leakage-controlled evaluation. The card states that UniProt, OAS and SCOP are licensed CC BY 4.0; cite Fournier et al., \"Protein Language Models: Is Scaling Necessary?\" (bioRxiv 2024) and the UniProt, OAS and SCOP papers.","displayTitle":"UR100P: AMPLIFY protein language model training and validation sets (UniProt, OAS paired antibodies, SCOP 2)","huggingFaceUrl":"https://huggingface.co/datasets/chandar-lab/UR100P","id":"3751dadd-0790-5ba2-b2ec-6f6e9186b18a","license":"CC-BY-4.0","licenseEvidenceUrl":"https://huggingface.co/datasets/chandar-lab/UR100P","objectCountEstimate":45,"probeStatus":"ok","referenceUrl":"https://doi.org/10.1101/2024.09.23.614603","region":"us","schemaCard":{"accessHints":[{"language":"python-datasets","snippet":"from datasets import load_dataset\nds = load_dataset(\"chandar-lab/UR100P\")"},{"language":"python-hub","snippet":"from huggingface_hub import snapshot_download\nsnapshot_download(repo_id=\"chandar-lab/UR100P\", repo_type=\"dataset\")"}],"caveats":["HuggingFace metadata: author=chandar-lab, license:cc-by-4.0, size_categories:100M<n<1B, format:csv, downloads=992"],"format":"mixed","generatedAt":"2026-11-17T14:22:59.424Z","layout":{"example":"OAS/test-00000-of-00001.csv","pattern":"{repoId}/{filename}"},"mountPath":"/mnt/archive","sampleHeaders":[{"bytes":"name,sequence\n>d0a00946-cbed-477c-b9b7-890e96030655|_A,QVQLQESGPGLVKPSGTLSLTCAVSGGSISSSNWWSWVRQPPGKGLEWIGEIYHSGSTNYNPSLKSRVTISVDKSKNQFSLKLSSVTAADTAVYYCARGSPTYDFWSAGREDGMDVWGQGTTVTVSS|DIQMTQSPSTLSASVGDRVTITCRASQSISSWLAWYQQKPGKAPKLLIYKASSLESGVPSRFSGSGSGTEFTLTISSLQPDDFATYYCQQYNSYPYTFGQGTKLEIK\n>d0a00946-cbed-477c-b9b7-890e96030655|_B,DIQMTQSPSTLSASVGDRVTITCRASQSISSWLAWYQQKPGKAPKLLIYKASSLESGVPSRFSGSGSGTEFTLTISSLQPDDFATYYCQQYNSYPYTFGQGTKLEIK|QVQLQESGPGLVKPSGTLSLTCAVSGGSISSSNWWSWVRQPPGKGLEWIGEIYHSGSTNYNPSLKSRVTISV","encoding":"utf8","key":"OAS/test-00000-of-00001.csv","truncatedAt":512}],"sampleObjects":["OAS/test-00000-of-00001.csv","OAS/train-00000-of-00001.csv","SCOP/test-00000-of-00001.csv","SCOP/train-00000-of-00001.csv","UniProt/test-00000-of-00001.csv"],"sidecars":[{"bytes":"---\nlicense: cc-by-4.0\ntags:\n- biology\npretty_name: UniRef100\nsize_categories:\n- 100M<n<1B\n---\n\n## UR100P\n\nThis dataset contains curated protein sequences from multiple sources and has been used to train and evaluate [AMPLIFY](https://huggingface.co/chandar-lab/AMPLIFY_350M), an efficient state-of-the-art protein language model. It combines data from [UniProt](https://www.uniprot.org), the [Observed Antibody Space](https://opig.stats.ox.ac.uk/webapps/oas/) (OAS), and the [Structural Classification of Proteins](https://www.ebi.ac.uk/pdbe/scop/) version 2 (SCOP 2) databases to enable task-specific validation of the models. All data were collected in December 2023, and sequences containing ambiguous amino acids (B, J, O, U, X, Z) were removed. For OAS, only paired heavy and light chain sequences were included, and the dataset was augmented by incorporating sequences in both heavy|light (Hc|Lc) and light|heavy (Lc|Hc) chain arrangements, separated by a chainbreak token `|`. MMseqs2 was used to filter out sequences in the train sets with >90% sequence identity to the validation sets, preventing data leakage and ensuring a fair evaluation. For more details, please refer to the [accompanying paper](https://www.biorxiv.org/content/10.1101/2024.09.23.614603v1).\n\n## Get Started\n\n```python\nfrom datasets import load_dataset\n\n# Load the entire dataset (all train and test sets)\n# dataset = load_dataset(\"chandar-lab/UR100P\")\n\n# Load all test sets\n# dataset = load_dataset(\"chandar-lab/UR100P\", split=\"test\")\n\n# Load the UniProt test set\ndataset = load_dataset(\"chandar-lab/UR100P\", data_dir=\"UniProt\", split=\"test\")\n```\n\n## Citations\n\nIf you find the models useful in your research, we ask that you cite the paper:\n\n```bibtex\n@article{Fournier2024.09.23.614603,\n\ttitle        = {Protein Language Models: Is Scaling Necessary?},\n\tauthor       = {Fournier, Quentin and Vernon, Robert M. and van der Sloot, Almer and Schulz, Benjamin and Chandar, Sarath and Langmead, Christopher James},\n\tyear         = {2024},\n\tjournal      = {bioRxiv},\n\tpublisher    = {Cold Spring Harbor Laboratory},\n\tdoi          = {10.1101/2024.09.23.614603},\n\turl          = {https://www.biorxiv.org/content/early/2024/09/23/2024.09.23.614603},\n\telocation-id = {2024.09.23.614603},\n\teprint       = {https://www.biorxiv.org/content/early/2024/09/23/2024.09.23.614603.full.pdf}\n}\n```\n\nIf you utilize UniProt, OAS, or SCOP, we ask that you cite their respective papers:\n\nThe UniProt Consortium, UniProt: the Universal Protein Knowledgebase in 2023, Nucleic Acids Research, Volume 51, Issue D1, 6 January 2023, Pages D523–D531, https://doi.org/10.1093/nar/gkac1052\n\nOlsen TH, Boyles F, Deane CM. Observed Antibody Space: A diverse database of cleaned, annotated, and translated unpaired and paired antibody sequences. Protein Science. 2022; 31: 141–146. https://doi.org/10.1002/pro.4205\n\nPDB-101: Educational resources supporting molecular explorations through biology and medicine. Christine Zardecki, Shuchismita Dutta, David S. Goodsell, Robert Lowe, Maria Voigt, Stephen K. Burley. (2022) Protein Science 31: 129-140 https://doi.org/10.1002/pro.4200\n\n## License\n\nUniProt, OAS, and SCOP are licensed under Creative Commons Attribution 4.0 International (CC BY 4.0).","encoding":"utf8","key":"README.md","name":"README.md","size":3248,"truncated":false}],"sourceObjectKey":"OAS/test-00000-of-00001.csv","starterCell":"import pandas as pd\ndf = pd.read_csv(\"hf://datasets/chandar-lab/UR100P/OAS/test-00000-of-00001.csv\")\ndf.head()","starterSamplePath":"OAS/test-00000-of-00001.csv","totalObjects":45},"summary":"The curated protein sequence corpus used to train and evaluate AMPLIFY, the efficient protein language model from Chandar Lab.","title":"UR100P: AMPLIFY protein language model training and validation sets (UniProt, OAS paired antibodies, SCOP 2)","updatedAt":"2026-11-17T20:21:07.787Z","visibility":"public","websiteUrl":"https://huggingface.co/datasets/chandar-lab/UR100P"}}