prima·bench
Repo ↗

Gene-level evals · results leaderboard

How models score on gene evals.

Every gene eval, scaffolded as a scored row — KPI tiles up top, then per-family score bars with a chance baseline. Numbers land once results.json is wired; until then the scaffold renders in its awaiting-results state.

42 evals · 4 families
01

Leaderboard

Awaiting results · prima-bench

Gene leaderboard

42 evals · 39 datasets · no models scored yet

Evals
42
graded
Datasets
39
sources
Models
pending
Best score
pending
No results.json wired up yet. The eval scaffold below is live; drop a document matching src/data/results.schema.md at src/data/results.json (or pass a results prop) and every bar fills in. Each track's tick marks the chance baseline.

Gene Function

29
EvalScoreawaiting
go_bpashburner_2000_nature_genetics/go_bp@0.0.2
fmax
go_bpashburner_2000_nature_genetics/go_bp@0.0.3
fmax
go_bp_discoveryashburner_2000_nature_genetics/go_bp_discovery@0.0.1
fmax
go_ccashburner_2000_nature_genetics/go_cc@0.0.2
fmax
go_ccashburner_2000_nature_genetics/go_cc@0.0.3
fmax
go_cc_discoveryashburner_2000_nature_genetics/go_cc_discovery@0.0.1
fmax
go_mfashburner_2000_nature_genetics/go_mf@0.0.2
fmax
go_mfashburner_2000_nature_genetics/go_mf@0.0.3
fmax
go_mf_discoveryashburner_2000_nature_genetics/go_mf_discovery@0.0.1
fmax
interpro_pfam_domainsblum_2025_nucleic_acids_research/interpro_pfam_domains@0.0.1
fmax
cell_type_markersborner_2021_nature_cell_biology/cell_type_markers@0.0.1
fmax
cell_type_markers_reference_alignedborner_2021_nature_cell_biology/cell_type_markers_reference_aligned@0.0.1
fmax
cell_type_markersfranzen_2019_database/cell_type_markers@0.0.1
fmax
cell_type_markers_reference_alignedfranzen_2019_database/cell_type_markers_reference_aligned@0.0.1
fmax
hpo_phenotypesgargano_2024_nucleic_acids_research/hpo_phenotypes@0.0.1
fmax
hpo_phenotypes_discoverygargano_2024_nucleic_acids_research/hpo_phenotypes_discovery@0.0.1
fmax
cell_type_markerskarlsson_2021_science_advances/cell_type_markers@0.0.1
fmax
cell_type_markers_reference_alignedkarlsson_2021_science_advances/cell_type_markers_reference_aligned@0.0.1
fmax
msigdb_gene_setsliberzon_2015_cell_systems/msigdb_gene_sets@0.0.1
fmax
reactome_pathwaysmilacic_2024_nucleic_acids_research/reactome_pathways@0.0.1
fmax
disease_associationsschriml_2025_nucleic_acids_research/disease_associations@0.0.1
fmax
disease_associations_discoveryschriml_2025_nucleic_acids_research/disease_associations_discovery@0.0.1
fmax
hgnc_gene_groupsseal_2023_nucleic_acids_research/hgnc_gene_groups@0.0.1
fmax
cell_type_markerstabula_sapiens_2022_science/cell_type_markers@0.0.1
fmax
cell_type_markers_reference_alignedtabula_sapiens_2022_science/cell_type_markers_reference_aligned@0.0.1
fmax
lineage_essentialitytsherniak_2017_cell/lineage_essentiality@0.0.1
fmax
lineage_essentiality_reference_alignedtsherniak_2017_cell/lineage_essentiality_reference_aligned@0.0.1
fmax
tissue_markersuhlen_2015_science/tissue_markers@0.0.1
fmax
tissue_markers_reference_aligneduhlen_2015_science/tissue_markers_reference_aligned@0.0.1
fmax

Gene Regression

5
EvalScoreawaiting
constraint_loeufchen_2024_nature/constraint_loeuf@0.0.1
spearman
dosage_phaplocollins_2022_cell/dosage_phaplo@0.0.1
spearman
dosage_ptriplocollins_2022_cell/dosage_ptriplo@0.0.1
spearman
haploinsufficiency_scoreshihab_2017_bioinformatics/haploinsufficiency_score@0.0.1
spearman
essentiality_gene_effecttsherniak_2017_cell/essentiality_gene_effect@0.0.1
spearman

Gene Classification

4
EvalScoreawaiting
humantfs_dbd_familylambert_2018_cell/humantfs_dbd_family@0.0.1
mcc
panther_familythomas_2022_protein_science/panther_family@0.0.1
mcc
panther_protein_classthomas_2022_protein_science/panther_protein_class@0.0.1
mcc
tfclass_familywingender_2018_nucleic_acids_research/tfclass_family@0.0.1
mcc

Sample Classification

4
EvalScoreawaiting
constraint_constrainedchen_2024_nature/constraint_constrained@0.0.1
sample_auroc
is_tflambert_2018_cell/is_tf@0.0.1
sample_auroc
dosage_haploinsufficiencyriggs_2012_clinical_genetics/dosage_haploinsufficiency@0.0.1
sample_auroc
dosage_triplosensitivityriggs_2012_clinical_genetics/dosage_triplosensitivity@0.0.1
sample_auroc
Scored with prima-bench · fill = metric value · tick = chance baseline · higher is better unless a metric is marked ↓.
02

References

The prima-artifacts prototypes this view rebuilds on our design tokens.