Table 1:

Ablation study and aggregated benchmark results for gRNAde. We report metrics averaged over 100 test sets samples and standard deviations across 3 consistent random seeds. The percentages reported in brackets for the 3D self-consistency scores are the percentage of designed samples within the ‘designability’ threshold values (scRMSD≤2Å, scTM≥0.45, scGDT≥0.5).

Split	Max. #states	Model	GNN	Max. train length	Perplexity (↓)	Native seq. recovery (↑)	Self-consistency metrics
							2D – EternaFold scMCC (↑)	scRMSD (↓)	3D – RhoFold scTM-score (↑)	scGDT_TS (↑)
Single-state split	1	AR	Equiv	500	1.77±0.07	0.438±0.01	0.624±0.07	13.01±1.18 (0.5%)	0.21±0.0 (14.3%)	0.22±0.0 (12.7%)
	1	AR	Equiv	1000	1.73±0.08	0.453±0.01	0.648±0.01	13.10±0.58 (1.0%)	0.20±0.0 (10.8%)	0.21±0.0 (10.6%)
	1	AR	Equiv	2500	1.41±0.01	0.513±0.01	0.633±0.03	11.76±0.91 (1.4%)	0.27±0.0 (28.8%)	0.27±0.0 (28.0%)
	1	AR	Equiv	5000	1.29±0.02	0.538±0.03	0.612±0.02	11.50±0.64 (1.9%)	0.28±0.0 (32.1%)	0.28±0.0 (26.2%)
	1	AR, rand	Equiv	5000	1.59±0.16	0.531±0.04	0.621±0.04	11.87±1.06 (1.9%)	0.26±0.0 (28.1%)	0.26±0.0 (24.1%)
	1	AR	Inv	5000	1.32±0.04	0.531±0.01	0.585±0.03	11.70±0.56 (1.3%)	0.26±0.0 (24.8%)	0.25±0.0 (20.1%)
	1	NAR	Inv	5000	1.54±0.04	0.571±0.00	0.430±0.02	14.26±0.51 (1.3%)	0.19±0.0 (15.9%)	0.18±0.0 (12.7%)
	1	NAR	Equiv	5000	1.46±0.06	0.584±0.00	0.473±0.02	13.04±0.88 (1.3%)	0.23±0.0 (24.0%)	0.22±0.0 (17.9%)
	3	AR	Equiv, DS	5000	1.23±0.05	0.539±0.01	0.620±0.01	11.47±1.05 (2.5%)	0.28±0.0 (31.4%)	0.28±0.0 (27.2%)
	5	AR	Equiv, DS	5000	1.25±0.01	0.539±0.02	0.596±0.03	11.90±1.00 (2.9%)	0.27±0.0 (31.6%)	0.26±0.0 (26.4%)
	Groundtruth sequence prediction baseline:				-	1.000±0.00	0.686±0.00	5.23±0.07 (27.9%)	0.56±0.0 (68.7%)	0.55±0.0 (68.7%)
	Random sequence prediction baseline:				-	0.251±0.00	0.012±0.00	24.40±0.34 (0.0%)	0.04±0.0 (0.0%)	0.02±0.0 (0.0%)
	ViennaRNA 2D-only baseline:				-	0.259±0.00	0.611±0.00	20.34±0.10 (0.0%)	0.07±0.0 (0.6%)	0.07±0.0 (1.1%)
Multi-state split	1	AR	Equiv	5000	1.51±0.01	0.481±0.00	0.573±0.04	21.83±0.53 (0.0%)	0.12±0.0 (2.6%)	0.15±0.0 (5.5%)
	3	AR	Equiv, DS	500	1.87±0.04	0.444±0.01	0.587±0.02	22.09±0.13 (0.0%)	0.12±0.0 (2.3%)	0.14±0.0 (5.7%)
	3	AR	Equiv, DS	1000	1.76±0.04	0.455±0.03	0.504±0.04	22.92±1.43 (0.0%)	0.11±0.0 (2.3%)	0.14±0.0 (5.8%)
	3	AR	Equiv, DS	2500	1.54±0.07	0.500±0.01	0.543±0.01	22.00±0.26 (0.0%)	0.11±0.0 (2.9%)	0.14±0.0 (3.7%)
	3	AR	Equiv, DS	5000	1.44±0.04	0.531±0.00	0.573±0.03	22.19±0.28 (0.0%)	0.12±0.0 (4.2%)	0.15±0.0 (7.5%)
	3	AR	Equiv, DSS	5000	1.37±0.04	0.540±0.03	0.574±0.03	22.20±0.43 (0.0%)	0.12±0.0 (4.0%)	0.15±0.0 (7.5%)
	5	AR	Equiv, DS	5000	1.37±0.03	0.510±0.00	0.514±0.00	21.80±0.08 (0.0%)	0.12±0.0 (2.9%)	0.14±0.0 (6.2%)
	1	NAR	Equiv	5000	1.81±0.03	0.489±0.00	0.372±0.03	24.18±0.63 (0.0%)	0.09±0.0 (2.2%)	0.12±0.0 (4.7%)
	3	NAR	Equiv, DS	5000	1.65±0.13	0.506±0.01	0.346±0.02	24.06±0.43 (0.0%)	0.08±0.0 (2.0%)	0.11±0.0 (2.9%)
	3	NAR	Equiv, DSS	5000	1.60±0.10	0.520±0.02	0.352±0.03	24.18±0.55 (0.0%)	0.09±0.0 (2.2%)	0.12±0.0 (4.7%)
	5	NAR	Equiv, DS	5000	1.59±0.21	0.517±0.01	0.339±0.01	24.16±0.75 (0.0%)	0.08±0.0 (2.2%)	0.10±0.0 (4.5%)
	Groundtruth sequence prediction baseline:				-	1.000±0.00	0.525±0.00	17.52±0.32 (3.9%)	0.25±0.0 (24.2%)	0.29±0.0 (31.4%)
	Random sequence prediction baseline:				-	0.249±0.00	0.013±0.00	31.00±0.20 (0.0%)	0.03±0.0 (0.0%)	0.02±0.0 (0.0%)
	ViennaRNA 2D-only baseline:				-	0.258±0.00	0.470±0.00	29.10±0.00 (0.0%)	0.05±0.0 (0.0%)	0.05±0.0 (0.0%)