{
  "@context": {
    "@language": "en",
    "@vocab": "https://schema.org/",
    "arrayShape": "cr:arrayShape",
    "citeAs": "cr:citeAs",
    "column": "cr:column",
    "conformsTo": "dct:conformsTo",
    "containedIn": "cr:containedIn",
    "cr": "http://mlcommons.org/croissant/",
    "data": {
      "@id": "cr:data",
      "@type": "@json"
    },
    "dataType": {
      "@id": "cr:dataType",
      "@type": "@vocab"
    },
    "dct": "http://purl.org/dc/terms/",
    "extract": "cr:extract",
    "field": "cr:field",
    "fileProperty": "cr:fileProperty",
    "fileObject": "cr:fileObject",
    "fileSet": "cr:fileSet",
    "format": "cr:format",
    "includes": "cr:includes",
    "isArray": "cr:isArray",
    "isLiveDataset": "cr:isLiveDataset",
    "jsonPath": "cr:jsonPath",
    "key": "cr:key",
    "md5": "cr:md5",
    "parentField": "cr:parentField",
    "path": "cr:path",
    "recordSet": "cr:recordSet",
    "references": "cr:references",
    "regex": "cr:regex",
    "repeated": "cr:repeated",
    "replace": "cr:replace",
    "sc": "https://schema.org/",
    "separator": "cr:separator",
    "source": "cr:source",
    "subField": "cr:subField",
    "transform": "cr:transform",
    "rai": "http://mlcommons.org/croissant/RAI/",
    "dataCollection": "rai:dataCollection",
    "dataCollectionType": "rai:dataCollectionType",
    "dataCollectionTimeframe": {
      "@id": "rai:dataCollectionTimeframe",
      "@type": "sc:Text"
    },
    "dataPreprocessingProtocol": "rai:dataPreprocessingProtocol",
    "dataAnnotationProtocol": "rai:dataAnnotationProtocol",
    "dataUseCases": "rai:dataUseCases",
    "dataBiases": "rai:dataBiases",
    "dataLimitations": "rai:dataLimitations",
    "dataSocialImpact": "rai:dataSocialImpact",
    "personalSensitiveInformation": "rai:personalSensitiveInformation",
    "dataReleaseMaintenancePlan": "rai:dataReleaseMaintenancePlan"
  },
  "@type": "sc:Dataset",
  "distribution": [
    {
      "@type": "cr:FileObject",
      "@id": "repo",
      "name": "repo",
      "description": "The Hugging Face git repository.",
      "contentUrl": "https://huggingface.co/datasets/lordim/BenchRep-T/tree/refs%2Fconvert%2Fparquet",
      "encodingFormat": "git+https",
      "sha256": "https://github.com/mlcommons/croissant/issues/80"
    },
    {
      "@type": "cr:FileObject",
      "@id": "repo-main",
      "name": "repo-main",
      "description": "The main branch of the Hugging Face git repository.",
      "contentUrl": "https://huggingface.co/datasets/lordim/BenchRep-T/tree/main",
      "encodingFormat": "git+https",
      "sha256": "https://github.com/mlcommons/croissant/issues/80"
    },
    {
      "@type": "cr:FileSet",
      "@id": "immuneml-repertoires",
      "name": "immuneml-repertoires",
      "description": "immuneML-ready AIRR Rearrangement repertoires, one gzipped TSV per specimen (see immuneML/README.md).",
      "containedIn": {
        "@id": "repo-main"
      },
      "encodingFormat": "application/gzip",
      "includes": "immuneML/**/repertoires/*.tsv.gz"
    },
    {
      "@type": "cr:FileSet",
      "@id": "immuneml-metadata",
      "name": "immuneml-metadata",
      "description": "immuneML metadata, task and fixed fold-split CSV files (see immuneML/README.md).",
      "containedIn": {
        "@id": "repo-main"
      },
      "encodingFormat": "text/csv",
      "includes": "immuneML/**/*.csv"
    }
  ],
  "recordSet": [],
  "conformsTo": "http://mlcommons.org/croissant/1.1",
  "name": "BenchRep-T",
  "description": "BenchRep-T is a benchmark for TCRβ (T-cell receptor β-chain) immune-repertoire classification. It bundles the multi-disease Mal-ID cohort together with five external disease cohorts (sequenced on the Adaptive Biotechnologies immunoSEQ platform: Savola_RA, Musvosvi_TB, Rawat_T1D, Mitchell_T1D, Emerson_CMV), providing per-specimen repertoire files, harmonized sample metadata, and auxiliary files used for the sequencing-depth scaling-law and driver-sequence experiments. Each specimen (one sequenced repertoire) is one example; the classification target is the specimen-level disease label. See the full description at https://huggingface.co/datasets/lordim/BenchRep-T.",
  "alternateName": [
    "lordim/BenchRep-T",
    "BenchRep-T"
  ],
  "creator": [
    {
      "@type": "Person",
      "name": "Alejandro Buendia"
    },
    {
      "@type": "Person",
      "name": "Chiho Im"
    },
    {
      "@type": "Person",
      "name": "Anshul Kundaje"
    },
    {
      "@type": "Person",
      "name": "Scott D Boyd"
    },
    {
      "@type": "Person",
      "name": "Liel Cohen-Lavi"
    }
  ],
  "citeAs": "@inproceedings{buendia2026benchrep,\n  title     = {BenchRep-T: A Systematic Evaluation of T-Cell Repertoire-Based Disease Diagnostics},\n  author    = {Buendia, Alejandro and Im, Chiho and Kundaje, Anshul and Boyd, Scott D and Cohen-Lavi, Liel},\n  booktitle = {Advances in Neural Information Processing Systems, Evaluations and Datasets Track},\n  year      = {2026}\n}",
  "keywords": [
    "mit",
    "1K<n<10K",
    "Region: US",
    "immunology",
    "t-cell-receptor",
    "tcr-beta",
    "immune-repertoire",
    "AIRR",
    "immunoseq",
    "repertoire-classification"
  ],
  "license": "https://choosealicense.com/licenses/mit/",
  "url": "https://huggingface.co/datasets/lordim/BenchRep-T",
  "dataCollection": "Bulk TCRβ immune-repertoire sequencing aggregated from previously published, de-identified human cohorts. Mal-ID is multi-disease B/T-cell receptor sequencing (Zaslavsky et al. 2025); the five external cohorts are Adaptive Biotechnologies immunoSEQ TCRβ repertoires from published disease studies: Savola et al. 2017 (rheumatoid arthritis), Musvosvi et al. 2023 (tuberculosis), Rawat et al. 2026 and Mitchell et al. 2022 (type 1 diabetes), and Emerson et al. 2017 (cytomegalovirus).",
  "dataCollectionType": "Secondary aggregation and harmonization of previously published, de-identified human immune-repertoire sequencing datasets.",
  "dataCollectionTimeframe": "Source studies published 2017–2026.",
  "dataPreprocessingProtocol": "Repertoire files were harmonized into per-specimen rearrangement tables (AIRR-style for Mal-ID; Adaptive immunoSEQ exports for the external cohorts) and gzip-compressed. Specimen-level disease labels and cross-validation fold assignments are provided in per-cohort metadata.tsv files. Specimens without a usable label were excluded (e.g., 66 BCR-only specimens dropped from Mal-ID; 25 unknown-CMV-status specimens dropped from Emerson_CMV).",
  "dataAnnotationProtocol": "Specimen-level disease/status labels are inherited from the clinical annotations of the original source studies; no additional manual annotation was performed for this benchmark.",
  "dataUseCases": "Intended for benchmarking machine-learning methods that classify disease or exposure status from an individual's TCRβ immune repertoire. Not intended for clinical diagnosis or individual-level medical decision-making.",
  "dataBiases": "Cohorts differ in demographic composition, recruitment, sequencing platform and depth, and case/control balance. Because labels come from heterogeneous studies, models may learn cohort- or platform-specific batch effects rather than disease biology, and performance may not transfer across cohorts or to underrepresented populations.",
  "dataLimitations": "Each cohort comprises a modest number of specimens (91–761). Only the TCRβ chain is provided (no paired α/β chains and no BCR). Mal-ID uses a different assay from the immunoSEQ external cohorts, limiting cross-cohort comparability. Demographic/clinical covariates are limited and vary by cohort.",
  "dataSocialImpact": "Repertoire-based classifiers could support non-invasive diagnostics, but deploying models that underperform on underrepresented demographic groups risks inequitable outcomes; this dataset is intended for research benchmarking only.",
  "personalSensitiveInformation": "Data consist of de-identified human immune-repertoire sequences plus a limited set of demographic and clinical fields (e.g., age, sex, ancestry/race, and HLA typing or autoantibody status for some cohorts). No direct identifiers are included. Note that immune-repertoire and HLA data can in principle be re-identifying; use in accordance with the data-use terms of the original source studies.",
  "dataReleaseMaintenancePlan": "Released as a static benchmark snapshot under the MIT License via the Hugging Face Hub; any updates are tracked through the repository commit history."
}