Function & pathways · gene eval

go_cc

GO term prediction (cellular component) — GO release 2026-05-19 (OBO) / GAF 2026-05-21, experimental-evidence only, terms with 15-500 annotated genes, paralog/gene-family-aware split (Ensembl Compara family ids). SUPERSEDED by 0.0.3 — labels are not true-path consistent: the recipe selected an upstream relation=='propagated' closure that was computed over the unfiltered corpus BEFORE the GENCODE v49 / IEA filters ran, so ~293k direct annotations were dropped and 29.77% of is_a edges have genes(child) not a subset of genes(parent). S-min here is weighted by marginal Resnik ic, whereas 0.0.3 weights it by conditional information accretion ia — a different measure, not on the same scale, so S-min values are not comparable across the two. Retained for provenance and for pinned adapters; use 0.0.3 for new work.

Function & pathways Term membership (multi-label) gene_function_classification

Spec

Target	GO term prediction (cellular component) — GO release 2026-05-19 (OBO) / GAF 2026-05-21, experimental-evidence only, terms with 15-500 annotated genes, paralog/gene-family-aware split (Ensembl Compara family ids). SUPERSEDED by 0.0.3 — labels are not true-path consistent: the recipe selected an upstream relation=='propagated' closure that was computed over the unfiltered corpus BEFORE the GENCODE v49 / IEA filters ran, so ~293k direct annotations were dropped and 29.77% of is_a edges have genes(child) not a subset of genes(parent). S-min here is weighted by marginal Resnik ic, whereas 0.0.3 weights it by conditional information accretion ia — a different measure, not on the same scale, so S-min values are not comparable across the two. Retained for provenance and for pinned adapters; use 0.0.3 for new work.
Axis	Function & pathways — Molecular function, biological process, and pathway membership.
Shape	Term membership (multi-label) · metric family: f_max
Task	gene_function_classification / gene_function
Dataset	ashburner_2000_nature_genetics/go_cc@0.0.2
Publication	ashburner_2000_nature_genetics
Metrics	gene_function_term_auprc gene_function_term_auroc gene_function_micro_auprc gene_function_fmax gene_function_smin
Temporal discovery	no — held-out genes
License	Declared per SPDX in the dataset's recipe README ↗ (commercial-use gated by CI).
References	doi:10.1038/75556 doi:10.1093/genetics/iyad031 doi:10.1093/nar/gku1113 linear:TECH-4594

IO contract

ic_column: ic
pairing_key: gene_term
required_metadata: gene_id, term
target_kind: binary

What the prediction looks like

This eval has the term membership (multi-label) shape. The figure is seeded to it — drag the control to watch the score respond.

Shape 1term membership · multi-labelFmax

A gene × term grid — which terms does the gene carry?

Used by GO · pathways · HPO · disease · domains · cell-type & tissue markers · lineage · temporal discovery

DNA repair

apoptosis

cell cycle

transcription

immune resp.

metabolism

TP53

BRCA1

MYC

CDK1

ALB

truth — gene really carries the term model calls it present (p ≥ threshold)cell shade = predicted probability

call present when p ≥ 0.50

Precision89%

Recall89%

F1 @ thr0.89

Fmax0.95

Truth is sparse — most cells are 0. The model emits a probability per cell (the shading); Fmax sweeps the threshold to the best-F1 operating point, so no single arbitrary cutoff is punished. Cell-type, lineage and temporal-discovery evals reuse this exact grid.

Run it

Fetch the censored splits, predict, score — task and dataset filled in.

import prima_bench as pb

# Downloads on demand into the local cache; returns local Paths.
train_path, test_path = pb.fetch_data(task="gene_function_classification", dataset="ashburner_2000_nature_genetics/go_cc@0.0.2")
train = pb.load(train_path, task="gene_function_classification", split="train", dataset="ashburner_2000_nature_genetics/go_cc@0.0.2")
test = pb.load(test_path, task="gene_function_classification", split="test", dataset="ashburner_2000_nature_genetics/go_cc@0.0.2")

# Your model produces predictions aligned to the eval input.
predictions = my_model(test)

# Score against the censored solution over this task's metric panel:
#   gene_function_term_auprc, gene_function_term_auroc, gene_function_micro_auprc, gene_function_fmax, gene_function_smin
run = pb.Evaluation(task="gene_function_classification", dataset="ashburner_2000_nature_genetics/go_cc@0.0.2")
run.update(predictions=predictions)   # one-shot; or stream shard-by-shard
result = run.compute()                # -> EvalResult
print(result)

Results

Awaiting scored models — the row scaffold is live; numbers land with results.json.

Awaiting results · prima-bench

go_cc · leaderboard

1 evals · 1 datasets · no models scored yet

Evals

graded

Datasets

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

Eval	Score	awaiting
go_cc`ashburner_2000_nature_genetics/go_cc@0.0.2`	—fmax	—