Tc-Ribo

Graph regression on RNA graphs, predicting one sequence-level expression shift induced by tetracycline binding. Evaluated by RMSE.

Dataset details

Task

Thermostable ribosome structure prediction (graph regression, RMSE metric). The primary catalog metric is RMSE.

Data

Thermostable ribosome structure prediction (graph regression, RMSE metric). The cataloged artifact reports 355 graphs.

Features

RNA samples include nucleotide identity, predicted or experimental 3D atomic coordinates as specified by the dataset, and task-specific scalar targets. Euclidean cutoff edges and spherical-harmonic neighborhoods are model construction, not additional measurements.

Splits and evaluation

The catalog records these protocols or variants: Standard split. Evaluation uses RMSE (lower is better). Exact masks or folds must come from the cited release.

Quirks and pitfalls

Predicted structures must not be described as experimental ground truth. Random sequence splits can place close motifs in train and test, circular angles require periodic losses, and experimental targets have assay-specific noise and missingness.

Sources

DualEquiNet RNA Benchmarks benchmark

SOTA Progression

new SOTA matches SOTA at that time other proposed result other baseline result

Leaderboard

#	Model	RMSE ↓	Paper
1	G-Heyna ★ proposed	0.5480 ± 0.0080	ICML 2025 Geometric Hyena Networks for Large-scale Equivariant Learning no code ICML 2025 May 2025 May '25
2	Equiformer GNN baseline	0.5500 ± 0.0090	ICML 2025 Geometric Hyena Networks for Large-scale Equivariant Learning no code ICML 2025 May 2025 May '25
3	G-Transformer GT baseline	0.5540 ± 0.0030	ICML 2025 Geometric Hyena Networks for Large-scale Equivariant Learning no code ICML 2025 May 2025 May '25
4	Transformer GT baseline	0.5560 ± 0.0010	ICML 2025 Geometric Hyena Networks for Large-scale Equivariant Learning no code ICML 2025 May 2025 May '25
5	Hyena Hyb ★ proposed	0.5600 ± 0.0020	ICML 2025 Geometric Hyena Networks for Large-scale Equivariant Learning no code ICML 2025 May 2025 May '25
6	DualEquiNet Hyb ★ proposed	0.6360 ± 0.0490	DualEquiNet RNA Benchmarks no code Jun 2025 Jun '25
7	SchNet GNN baseline	0.6910 ± 0.0180	ICML 2025 Geometric Hyena Networks for Large-scale Equivariant Learning no code ICML 2025 May 2025 May '25
8	FastEGNN GNN baseline	0.7040 ± 0.0050	ICML 2025 Geometric Hyena Networks for Large-scale Equivariant Learning no code ICML 2025 May 2025 May '25
9	TFN GNN baseline	0.7100 ± 0.0090	ICML 2025 Geometric Hyena Networks for Large-scale Equivariant Learning no code ICML 2025 May 2025 May '25
10	GotenNet GNN baseline	0.7180 ± 0.0020	DualEquiNet RNA Benchmarks no code Jun 2025 Jun '25
11	EGNN GNN baseline	0.7280 ± 0.0010	ICML 2025 Geometric Hyena Networks for Large-scale Equivariant Learning no code ICML 2025 May 2025 May '25
12	HEGNN GNN baseline	0.7290 ± 0.0070	DualEquiNet RNA Benchmarks no code Jun 2025 Jun '25
13	FastEGNN GNN baseline	0.7310 ± 0.0060	DualEquiNet RNA Benchmarks no code Jun 2025 Jun '25
14	GVPGNN GNN baseline	0.7360 ± 0.0030	DualEquiNet RNA Benchmarks no code Jun 2025 Jun '25
15	SchNet GNN baseline	0.7370 ± 0.0020	DualEquiNet RNA Benchmarks no code Jun 2025 Jun '25
16	LEFTNet GNN baseline	0.7490 ± 0.0060	ICML 2025 Geometric Hyena Networks for Large-scale Equivariant Learning no code ICML 2025 May 2025 May '25
17	TMD-ET GNN baseline	0.7500 ± 0.0040	ICML 2025 Geometric Hyena Networks for Large-scale Equivariant Learning no code ICML 2025 May 2025 May '25

SOTA performance set a new SOTA at the time matched the SOTA at the time

Structure

Coverage