{
  "schema_version": 1,
  "project": "GenomicTouchstone",
  "scope_as_of": "2026-10-18",
  "description": "Task catalogue for the revised benchmark scope. Dataset settings include related sources, lengths and split protocols; they do not represent 103 independent datasets.",
  "counts": {
    "task_types": 44,
    "dataset_settings": 103,
    "main_settings": 100,
    "long_range_settings": 3,
    "settings_by_modality": {
      "DNA": 55,
      "RNA": 33,
      "Protein": 15
    }
  },
  "tasks": [
    {
      "name": "BRCA1/2 SNV Functional Impact Classification",
      "modality": "DNA",
      "category": "Genetic Variant Effect Prediction",
      "settings": [
        {
          "id": "brca1_1024_coding",
          "objective": "Classification",
          "scope": "main"
        },
        {
          "id": "brca1_1024_noncoding",
          "objective": "Classification",
          "scope": "main"
        },
        {
          "id": "brca2_1024_coding",
          "objective": "Classification",
          "scope": "main"
        },
        {
          "id": "brca2_1024_noncoding",
          "objective": "Classification",
          "scope": "main"
        }
      ]
    },
    {
      "name": "Genetic Disease Classification",
      "modality": "DNA",
      "category": "Genetic Variant Effect Prediction",
      "settings": [
        {
          "id": "breast_disease_type_classification_512",
          "objective": "Classification",
          "scope": "main"
        },
        {
          "id": "cardiovascular_disorders_classification_512",
          "objective": "Classification",
          "scope": "main"
        },
        {
          "id": "kidney_disease_type_classification_512",
          "objective": "Classification",
          "scope": "main"
        }
      ]
    },
    {
      "name": "Pathogenicity Classification",
      "modality": "DNA",
      "category": "Genetic Variant Effect Prediction",
      "settings": [
        {
          "id": "pathogenicity_classification_512",
          "objective": "Classification",
          "scope": "main"
        }
      ]
    },
    {
      "name": "Splice Variant Effect Prediction",
      "modality": "DNA",
      "category": "Genetic Variant Effect Prediction",
      "settings": [
        {
          "id": "splicevardb_1024_exonic",
          "objective": "Classification",
          "scope": "main"
        },
        {
          "id": "splicevardb_1024_intronic",
          "objective": "Classification",
          "scope": "main"
        }
      ]
    },
    {
      "name": "eQTL Prediction",
      "modality": "DNA",
      "category": "Genetic Variant Effect Prediction",
      "settings": [
        {
          "id": "eqtl_whole_blood_dnalongbench",
          "objective": "Classification",
          "scope": "long_range"
        }
      ]
    },
    {
      "name": "CpG Methylation Prediction (Classification)",
      "modality": "DNA",
      "category": "Genomic Function Annotation",
      "settings": [
        {
          "id": "cpg_methylation",
          "objective": "Classification",
          "scope": "main"
        }
      ]
    },
    {
      "name": "CpG Methylation Prediction (Regression)",
      "modality": "DNA",
      "category": "Genomic Function Annotation",
      "settings": [
        {
          "id": "cpg_methylation_regression",
          "objective": "Regression",
          "scope": "main"
        }
      ]
    },
    {
      "name": "Enhancer Types Classification",
      "modality": "DNA",
      "category": "Genomic Function Annotation",
      "settings": [
        {
          "id": "enhancers_types",
          "objective": "Classification",
          "scope": "main"
        }
      ]
    },
    {
      "name": "Epigenetic Marks Prediction (Classification)",
      "modality": "DNA",
      "category": "Genomic Function Annotation",
      "settings": [
        {
          "id": "histone_annotation_1",
          "objective": "Classification",
          "scope": "main"
        }
      ]
    },
    {
      "name": "Epigenetic Marks Prediction (Regression)",
      "modality": "DNA",
      "category": "Genomic Function Annotation",
      "settings": [
        {
          "id": "histone_annotation_regression",
          "objective": "Regression",
          "scope": "main"
        }
      ]
    },
    {
      "name": "Promoter Annotation",
      "modality": "DNA",
      "category": "Genomic Function Annotation",
      "settings": [
        {
          "id": "core_promoter_annotation",
          "objective": "Classification",
          "scope": "main"
        },
        {
          "id": "promoter_annotation",
          "objective": "Classification",
          "scope": "main"
        }
      ]
    },
    {
      "name": "Regulation Element Prediction",
      "modality": "DNA",
      "category": "Genomic Function Annotation",
      "settings": [
        {
          "id": "regulation",
          "objective": "Classification",
          "scope": "main"
        },
        {
          "id": "regulation_multilabel",
          "objective": "Classification",
          "scope": "main"
        }
      ]
    },
    {
      "name": "Species Classification",
      "modality": "DNA",
      "category": "Genomic Function Annotation",
      "settings": [
        {
          "id": "species_classification_1024",
          "objective": "Classification",
          "scope": "main"
        },
        {
          "id": "species_classification_128",
          "objective": "Classification",
          "scope": "main"
        },
        {
          "id": "species_classification_2048",
          "objective": "Classification",
          "scope": "main"
        },
        {
          "id": "species_classification_256",
          "objective": "Classification",
          "scope": "main"
        },
        {
          "id": "species_classification_512",
          "objective": "Classification",
          "scope": "main"
        }
      ]
    },
    {
      "name": "Splice Sites Prediction",
      "modality": "DNA",
      "category": "Genomic Function Annotation",
      "settings": [
        {
          "id": "splice_sites_acceptor",
          "objective": "Classification",
          "scope": "main"
        },
        {
          "id": "splice_sites_donor",
          "objective": "Classification",
          "scope": "main"
        },
        {
          "id": "splice_sites_all",
          "objective": "Classification",
          "scope": "main"
        }
      ]
    },
    {
      "name": "Enhancer Activity Prediction",
      "modality": "DNA",
      "category": "Regulatory Mechanism Modeling",
      "settings": [
        {
          "id": "enhancer_activity_deepstarr_hct116",
          "objective": "Regression",
          "scope": "main"
        },
        {
          "id": "enhancer_activity_hct116_brd2",
          "objective": "Regression",
          "scope": "main"
        },
        {
          "id": "enhancer_activity_hct116_brd4",
          "objective": "Regression",
          "scope": "main"
        },
        {
          "id": "enhancer_activity_hct116_cdk7",
          "objective": "Regression",
          "scope": "main"
        },
        {
          "id": "enhancer_activity_hct116_cdk9",
          "objective": "Regression",
          "scope": "main"
        },
        {
          "id": "enhancer_activity_hct116_med14",
          "objective": "Regression",
          "scope": "main"
        },
        {
          "id": "enhancer_activity_hct116_p300cbp",
          "objective": "Regression",
          "scope": "main"
        },
        {
          "id": "enhancer_activity_hct116_parental",
          "objective": "Regression",
          "scope": "main"
        }
      ]
    },
    {
      "name": "Enhancer Promoter Interaction Prediction",
      "modality": "DNA",
      "category": "Regulatory Mechanism Modeling",
      "settings": [
        {
          "id": "enhancer_promoter_interaction_gm12878",
          "objective": "Classification",
          "scope": "main"
        },
        {
          "id": "enhancer_promoter_interaction_hela_s3",
          "objective": "Classification",
          "scope": "main"
        },
        {
          "id": "enhancer_promoter_interaction_huvec",
          "objective": "Classification",
          "scope": "main"
        },
        {
          "id": "enhancer_promoter_interaction_imr90",
          "objective": "Classification",
          "scope": "main"
        },
        {
          "id": "enhancer_promoter_interaction_k562",
          "objective": "Classification",
          "scope": "main"
        },
        {
          "id": "enhancer_promoter_interaction_nhek",
          "objective": "Classification",
          "scope": "main"
        },
        {
          "id": "enhancer_promoter_interaction_gasperini_k562",
          "objective": "Classification",
          "scope": "main"
        }
      ]
    },
    {
      "name": "Enhancer Target Gene Prediction",
      "modality": "DNA",
      "category": "Regulatory Mechanism Modeling",
      "settings": [
        {
          "id": "enhancer_target_gene_gasperini_k562",
          "objective": "Classification",
          "scope": "long_range"
        }
      ]
    },
    {
      "name": "Gene Expression Prediction",
      "modality": "DNA",
      "category": "Regulatory Mechanism Modeling",
      "settings": [
        {
          "id": "bulk_rna_expression_1024",
          "objective": "Regression",
          "scope": "main"
        },
        {
          "id": "bulk_rna_expression_128",
          "objective": "Regression",
          "scope": "main"
        },
        {
          "id": "bulk_rna_expression_2048",
          "objective": "Regression",
          "scope": "main"
        },
        {
          "id": "bulk_rna_expression_256",
          "objective": "Regression",
          "scope": "main"
        },
        {
          "id": "bulk_rna_expression_512",
          "objective": "Regression",
          "scope": "main"
        }
      ]
    },
    {
      "name": "Regulatory Activity Prediction",
      "modality": "DNA",
      "category": "Regulatory Mechanism Modeling",
      "settings": [
        {
          "id": "regulatory_activity_gm12878",
          "objective": "Regression",
          "scope": "main"
        },
        {
          "id": "regulatory_activity_h1esc",
          "objective": "Regression",
          "scope": "main"
        },
        {
          "id": "regulatory_activity_hepg2",
          "objective": "Regression",
          "scope": "main"
        },
        {
          "id": "regulatory_activity_imr90",
          "objective": "Regression",
          "scope": "main"
        },
        {
          "id": "regulatory_activity_k562",
          "objective": "Regression",
          "scope": "main"
        }
      ]
    },
    {
      "name": "Transcription Initiation Signal Prediction (Long-range)",
      "modality": "DNA",
      "category": "Regulatory Mechanism Modeling",
      "settings": [
        {
          "id": "transcription_initiation_signal_prediction",
          "objective": "Regression",
          "scope": "long_range"
        }
      ]
    },
    {
      "name": "APA Isoform Prediction",
      "modality": "RNA",
      "category": "Post-transcriptional Regulation",
      "settings": [
        {
          "id": "isoform",
          "objective": "Regression",
          "scope": "main"
        }
      ]
    },
    {
      "name": "Modification Prediction",
      "modality": "RNA",
      "category": "Post-transcriptional Regulation",
      "settings": [
        {
          "id": "modification",
          "objective": "Classification",
          "scope": "main"
        }
      ]
    },
    {
      "name": "SARS-CoV-2 Vaccine Degradation Prediction",
      "modality": "RNA",
      "category": "Post-transcriptional Regulation",
      "settings": [
        {
          "id": "cov_vaccine_degradation",
          "objective": "Regression",
          "scope": "main"
        },
        {
          "id": "cov_vaccine_degradation_random_iid",
          "objective": "Regression",
          "scope": "main"
        }
      ]
    },
    {
      "name": "Programmable RNA Switches Prediction",
      "modality": "RNA",
      "category": "RNA Engineering Applications",
      "settings": [
        {
          "id": "programmable_rna_switches",
          "objective": "Regression",
          "scope": "main"
        }
      ]
    },
    {
      "name": "Tc-Riboswitches Prediction",
      "modality": "RNA",
      "category": "RNA Engineering Applications",
      "settings": [
        {
          "id": "tc_riboswitches",
          "objective": "Regression",
          "scope": "main"
        },
        {
          "id": "tc_riboswitches_random_iid",
          "objective": "Regression",
          "scope": "main"
        }
      ]
    },
    {
      "name": "mRFP Expression Prediction",
      "modality": "RNA",
      "category": "RNA Engineering Applications",
      "settings": [
        {
          "id": "mrfp_expression",
          "objective": "Regression",
          "scope": "main"
        }
      ]
    },
    {
      "name": "Expression Level Prediction",
      "modality": "RNA",
      "category": "RNA Functional Studies",
      "settings": [
        {
          "id": "hek_el",
          "objective": "Regression",
          "scope": "main"
        },
        {
          "id": "muscle_el",
          "objective": "Regression",
          "scope": "main"
        },
        {
          "id": "pc3_el",
          "objective": "Regression",
          "scope": "main"
        }
      ]
    },
    {
      "name": "Mean Ribosome Loading Prediction",
      "modality": "RNA",
      "category": "RNA Functional Studies",
      "settings": [
        {
          "id": "mean_ribosome_loading",
          "objective": "Regression",
          "scope": "main"
        },
        {
          "id": "mean_ribosome_loading_human",
          "objective": "Regression",
          "scope": "main"
        }
      ]
    },
    {
      "name": "Non-coding RNA Classification",
      "modality": "RNA",
      "category": "RNA Functional Studies",
      "settings": [
        {
          "id": "noncodingrnafamily",
          "objective": "Classification",
          "scope": "main"
        }
      ]
    },
    {
      "name": "RNA Localization Prediction",
      "modality": "RNA",
      "category": "RNA Functional Studies",
      "settings": [
        {
          "id": "rna_localization_rnagps_apexseq",
          "objective": "Classification",
          "scope": "main"
        }
      ]
    },
    {
      "name": "Transcript and Protein Abundance Prediction",
      "modality": "RNA",
      "category": "RNA Functional Studies",
      "settings": [
        {
          "id": "pa_athaliana",
          "objective": "Regression",
          "scope": "main"
        },
        {
          "id": "pa_dmelanogaster",
          "objective": "Regression",
          "scope": "main"
        },
        {
          "id": "pa_ecoli",
          "objective": "Regression",
          "scope": "main"
        },
        {
          "id": "pa_hsapiens",
          "objective": "Regression",
          "scope": "main"
        },
        {
          "id": "pa_scerevisiae",
          "objective": "Regression",
          "scope": "main"
        },
        {
          "id": "ta_athaliana",
          "objective": "Regression",
          "scope": "main"
        },
        {
          "id": "ta_dmelanogaster",
          "objective": "Regression",
          "scope": "main"
        },
        {
          "id": "ta_ecoli",
          "objective": "Regression",
          "scope": "main"
        },
        {
          "id": "ta_hsapiens",
          "objective": "Regression",
          "scope": "main"
        },
        {
          "id": "ta_hvolcanii",
          "objective": "Regression",
          "scope": "main"
        },
        {
          "id": "ta_ppastoris",
          "objective": "Regression",
          "scope": "main"
        },
        {
          "id": "ta_scerevisiae",
          "objective": "Regression",
          "scope": "main"
        }
      ]
    },
    {
      "name": "Translation Efficiency Prediction",
      "modality": "RNA",
      "category": "RNA Functional Studies",
      "settings": [
        {
          "id": "hek_te",
          "objective": "Regression",
          "scope": "main"
        },
        {
          "id": "muscle_te",
          "objective": "Regression",
          "scope": "main"
        },
        {
          "id": "pc3_te",
          "objective": "Regression",
          "scope": "main"
        },
        {
          "id": "translation_efficiency_ribonn_human_mean",
          "objective": "Regression",
          "scope": "main"
        }
      ]
    },
    {
      "name": "mRNA Stability Prediction",
      "modality": "RNA",
      "category": "RNA Functional Studies",
      "settings": [
        {
          "id": "mrna_stability",
          "objective": "Regression",
          "scope": "main"
        },
        {
          "id": "mrna_half_life_saluki_human",
          "objective": "Regression",
          "scope": "main"
        }
      ]
    },
    {
      "name": "Enzyme Classification",
      "modality": "Protein",
      "category": "Protein Functional Annotation",
      "settings": [
        {
          "id": "enzyme",
          "objective": "Classification",
          "scope": "main"
        }
      ]
    },
    {
      "name": "Post-translational Modifications Prediction",
      "modality": "Protein",
      "category": "Protein Functional Annotation",
      "settings": [
        {
          "id": "ptm",
          "objective": "Classification",
          "scope": "main"
        }
      ]
    },
    {
      "name": "Protein Domain Classification",
      "modality": "Protein",
      "category": "Protein Functional Annotation",
      "settings": [
        {
          "id": "domain",
          "objective": "Classification",
          "scope": "main"
        }
      ]
    },
    {
      "name": "Protein Subcellular Localization (DeepLoc2)",
      "modality": "Protein",
      "category": "Protein Functional Annotation",
      "settings": [
        {
          "id": "protein_localization_deeploc2",
          "objective": "Classification",
          "scope": "main"
        }
      ]
    },
    {
      "name": "Protein Variant Classification",
      "modality": "Protein",
      "category": "Protein Functional Annotation",
      "settings": [
        {
          "id": "variant",
          "objective": "Classification",
          "scope": "main"
        }
      ]
    },
    {
      "name": "Beta-Lactamase Activity Prediction",
      "modality": "Protein",
      "category": "Protein Property Prediction",
      "settings": [
        {
          "id": "beta_lactamase_complete",
          "objective": "Regression",
          "scope": "main"
        },
        {
          "id": "beta_lactamase_unique",
          "objective": "Regression",
          "scope": "main"
        }
      ]
    },
    {
      "name": "Fluorescence Prediction",
      "modality": "Protein",
      "category": "Protein Property Prediction",
      "settings": [
        {
          "id": "fluorescence_amacgfp",
          "objective": "Regression",
          "scope": "main"
        },
        {
          "id": "fluorescence_avgfp",
          "objective": "Regression",
          "scope": "main"
        },
        {
          "id": "fluorescence_cgregfp",
          "objective": "Regression",
          "scope": "main"
        },
        {
          "id": "fluorescence_pplugfp2",
          "objective": "Regression",
          "scope": "main"
        }
      ]
    },
    {
      "name": "Melting Point Prediction",
      "modality": "Protein",
      "category": "Protein Property Prediction",
      "settings": [
        {
          "id": "melting_point",
          "objective": "Regression",
          "scope": "main"
        }
      ]
    },
    {
      "name": "Protein Stability Prediction",
      "modality": "Protein",
      "category": "Protein Property Prediction",
      "settings": [
        {
          "id": "stability",
          "objective": "Regression",
          "scope": "main"
        }
      ]
    },
    {
      "name": "Secondary Structure Analysis",
      "modality": "Protein",
      "category": "Structural Analysis and Prediction",
      "settings": [
        {
          "id": "structure",
          "objective": "Classification",
          "scope": "main"
        }
      ]
    },
    {
      "name": "Transmembrane Region Prediction",
      "modality": "Protein",
      "category": "Structural Analysis and Prediction",
      "settings": [
        {
          "id": "transmembrane",
          "objective": "Classification",
          "scope": "main"
        }
      ]
    }
  ]
}