Virtual Embryo Three tasks, one shared resource
Validation and test ground truth are both withheld. A validation submission returns a leaderboard score, not the answers — those are released only when the final test set is.
Temporal gene-expression distribution prediction
Given stages before a target time, predict the gene-expression distribution — a set of cells — at a future stage. Tests temporal extrapolation.
- Modality
- single-cell RNA
- Predicts
- expression
- Splits
- train E8.5 · E9.5 → val E10.5 → test E12.5
- Scored on
- DE gene recovery (25%), Change direction (25%), Cell-state distribution (30%), Gene-gene co-variation (20%)
- Why it’s hard
- No observed stage sits between E10.5 and E12.5, so temporal interpolation has nothing to lean on; and only about a third of target cells have a same-type predecessor to extrapolate a velocity from.
Spatial-temporal multiscale prediction
Predict future gene expression and 3D spatial location jointly, across molecular, cellular and spatial scales. Evaluated at two organ scopes.
- Modality
- 3D MERFISH (500-gene panel)
- Predicts
- expression + 3D coordinates
- Splits
- heart: val E8.5, E10.5 / test E8.75, E12.5 · embryo: val E7.5 / test E7.75
- Scored on
- Expression change (25%), Cell-state distribution (25%), Tissue shape and growth scale (25%), Local spatial organisation (25%)
- Why it’s hard
- Getting the marginals right is not enough — tissue geometry carries a quarter of the score on its own. And the held-out stages sit both inside the observed range and beyond it, so a model has to interpolate and extrapolate.
Mutant perturbation prediction
Predict a held-out knockout — expression and 3D coordinates — from wild-type development plus one observed perturbation.
- Modality
- 3D MERFISH (500-gene panel)
- Predicts
- expression + 3D coordinates
- Splits
- train Mab21l2 (E9.5) → val Gata4 (E8.75) → test β-catenin (E8.75)
- Scored on
- Response gene recovery (30%), Response direction (25%), Response magnitude (25%), Cell-state distribution (20%)
- Why it’s hard
- Scored on the perturbation effect (WT→KO delta), not absolute expression: doing nothing already scores 0.956 absolute pseudobulk Pearson, so only the change counts.
Two parallel tracks
Both tracks address the same three tasks and are scored on the same metrics and hidden test sets. Prizes are awarded separately, so the leaderboards directly contrast human-designed and agent-designed approaches.
- Track 1Human Team
- Conventional ML-competition workflow.
Methods designed and supervised by human participants. Algorithm/model design → submission → evaluation. Standard NeurIPS competition track.
- Track 2Agent Team
- Coding agents / LLM-driven recursive systems.
Methods produced by coding agents or LLM-based evolutionary algorithms that iteratively select, mutate, and evaluate candidate solutions without human supervision. The agent system and the complete evolutionary trace must be shared before prize evaluation.