EXPERIMENT 03 / SINGLE-TARGET
DUDE-Z: 43-target search
Search the DUDE-Z 43-target benchmark for the strongest docking candidates, treating each protein as an independent search. Explore selected target results below.
- Library
- 1,167,366 molecules
- Paid budget
- 116,946 measurements / target
- Targets
- 43-target benchmark · selected targets below
- Search seeds
- 100
THE EXPERIMENT
What are we looking for?
For each protein, find the molecules with the lowest recorded DOCK score. The displayed results cover AA2AR, HDAC8, ACES, CSF1R, and XIAP, with equal-weight means across these targets.
Minimize the raw docking score separately for each target
The DUDE-Z Goldilocks library uses recorded DOCK scores. Comparisons use current canonical membership while preserving the reference search’s original duplicate and unavailable-call charges.
FINAL QUALITY
Results at the full budget
Means give equal weight to the 5 displayed targets at the same per-target budget. Sample SD describes target-to-target variation; each target uses seed 100. Bold marks the higher mean.
| Metric | MolScout | RAD | Difference | Ratio |
|---|---|---|---|---|
| Recall@100 | 89.00%± 8.97 pp SD | 85.60%± 10.81 pp SD | +3.40 pp | 1.04× |
| Recall@250 | 91.12%± 4.50 pp SD | 85.60%± 6.51 pp SD | +5.52 pp | 1.06× |
| Recall@500 | 90.80%± 3.73 pp SD | 84.76%± 4.98 pp SD | +6.04 pp | 1.07× |
| Recall@1000 | 90.48%± 3.29 pp SD | 83.34%± 4.31 pp SD | +7.14 pp | 1.09× |
| AUC@100 | 0.815± 0.116 SD | 0.723± 0.147 SD | +0.092 | 1.13× |
| AUC@250 | 0.818± 0.072 SD | 0.703± 0.105 SD | +0.115 | 1.16× |
| AUC@500 | 0.805± 0.073 SD | 0.687± 0.091 SD | +0.118 | 1.17× |
| AUC@1000 | 0.786± 0.064 SD | 0.654± 0.078 SD | +0.132 | 1.20× |
Mean discovery curves
Compare the methods for the top 100, 250, 500, or 1,000 molecules.
Recall
Percentage of the fixed top K molecules discovered with all required measurements. Higher is better.
Top 100
Top 250
Top 500
Top 1,000
AUC
Average fractional Recall@K over the measurements made so far, from 0 to 1. Higher values mean earlier discovery.
Top 100
Top 250
Top 500
Top 1,000
INDIVIDUAL RESULTS
Individual target results
Each run uses 116,946 charged measurements per target. Bold marks the higher value; higher is better for both metrics.
| Top K | MolScout | RAD |
|---|---|---|
| 100 | 91.00% | 87.00% |
| 250 | 91.20% | 81.20% |
| 500 | 88.60% | 80.80% |
| 1000 | 88.00% | 80.30% |
| Top K | MolScout | RAD |
|---|---|---|
| 100 | 97.00% | 98.00% |
| 250 | 95.60% | 95.20% |
| 500 | 93.20% | 92.20% |
| 1000 | 89.80% | 89.60% |
| Top K | MolScout | RAD |
|---|---|---|
| 100 | 96.00% | 91.00% |
| 250 | 94.80% | 88.00% |
| 500 | 96.20% | 87.40% |
| 1000 | 96.20% | 86.10% |
| Top K | MolScout | RAD |
|---|---|---|
| 100 | 86.00% | 83.00% |
| 250 | 89.60% | 85.20% |
| 500 | 88.40% | 82.80% |
| 1000 | 88.60% | 80.70% |
| Top K | MolScout | RAD |
|---|---|---|
| 100 | 75.00% | 69.00% |
| 250 | 84.40% | 78.40% |
| 500 | 87.60% | 80.60% |
| 1000 | 89.80% | 80.00% |
| Top K | MolScout | RAD |
|---|---|---|
| 100 | 0.801 | 0.647 |
| 250 | 0.777 | 0.613 |
| 500 | 0.748 | 0.601 |
| 1000 | 0.730 | 0.574 |
| Top K | MolScout | RAD |
|---|---|---|
| 100 | 0.939 | 0.925 |
| 250 | 0.900 | 0.862 |
| 500 | 0.865 | 0.825 |
| 1000 | 0.805 | 0.763 |
| Top K | MolScout | RAD |
|---|---|---|
| 100 | 0.928 | 0.826 |
| 250 | 0.894 | 0.756 |
| 500 | 0.896 | 0.733 |
| 1000 | 0.883 | 0.706 |
| Top K | MolScout | RAD |
|---|---|---|
| 100 | 0.718 | 0.644 |
| 250 | 0.754 | 0.667 |
| 500 | 0.728 | 0.632 |
| 1000 | 0.726 | 0.609 |
| Top K | MolScout | RAD |
|---|---|---|
| 100 | 0.690 | 0.571 |
| 250 | 0.765 | 0.619 |
| 500 | 0.790 | 0.642 |
| 1000 | 0.788 | 0.620 |
MEASUREMENT EFFICIENCY
Budget to match the reference
Each target level is the reference method’s final mean. The recorded crossing is the first budget that reaches that level. Budget reduction compares MolScout’s crossing with the full measurement budget.
| Metric | Reference final level | MolScout crossing | Reference first crossing | Full budget | Budget reduction |
|---|---|---|---|---|---|
| Recall@100 | 85.60% | 64,385 | 115,602 | 116,946 | 44.9% |
| Recall@250 | 85.60% | 53,957 | 116,700 | 116,946 | 53.9% |
| Recall@500 | 84.76% | 55,278 | 116,700 | 116,946 | 52.7% |
| Recall@1000 | 83.34% | 56,164 | 116,744 | 116,946 | 52.0% |
Crossing budgets by target
| Metric | Reference final level | MolScout crossing | Reference first crossing | Full budget | Budget reduction |
|---|---|---|---|---|---|
| AA2AR · Recall@100 | 87.00% | 69,143 | 115,602 | 116,946 | 40.9% |
| AA2AR · Recall@250 | 81.20% | 48,971 | 115,602 | 116,946 | 58.1% |
| AA2AR · Recall@500 | 80.80% | 64,559 | 115,602 | 116,946 | 44.8% |
| AA2AR · Recall@1000 | 80.30% | 64,563 | 116,411 | 116,946 | 44.8% |
| HDAC8 · Recall@100 | 98.00% | Not reached | 62,505 | 116,946 | — |
| HDAC8 · Recall@250 | 95.20% | 99,057 | 114,799 | 116,946 | 15.3% |
| HDAC8 · Recall@500 | 92.20% | 89,248 | 114,799 | 116,946 | 23.7% |
| HDAC8 · Recall@1000 | 89.60% | 111,771 | 116,730 | 116,946 | 4.4% |
| ACES · Recall@100 | 91.00% | 11,810 | 91,703 | 116,946 | 89.9% |
| ACES · Recall@250 | 88.00% | 19,917 | 116,700 | 116,946 | 83.0% |
| ACES · Recall@500 | 87.40% | 23,969 | 116,700 | 116,946 | 79.5% |
| ACES · Recall@1000 | 86.10% | 25,007 | 116,700 | 116,946 | 78.6% |
| CSF1R · Recall@100 | 83.00% | 77,749 | 113,644 | 116,946 | 33.5% |
| CSF1R · Recall@250 | 85.20% | 80,027 | 113,644 | 116,946 | 31.6% |
| CSF1R · Recall@500 | 82.80% | 83,207 | 113,644 | 116,946 | 28.9% |
| CSF1R · Recall@1000 | 80.70% | 69,111 | 113,999 | 116,946 | 40.9% |
| XIAP · Recall@100 | 69.00% | 23,760 | 109,972 | 116,946 | 79.7% |
| XIAP · Recall@250 | 78.40% | 37,963 | 109,972 | 116,946 | 67.5% |
| XIAP · Recall@500 | 80.60% | 39,384 | 113,825 | 116,946 | 66.3% |
| XIAP · Recall@1000 | 80.00% | 40,258 | 116,744 | 116,946 | 65.6% |
HOW WE COUNT
Measurements, recall, and AUC
- One measurement
- One charged molecule–target score. Initialization, repeated calls, and unavailable evaluations keep the charges recorded by the experiment.
- Recall@K
- The fraction of the fixed top-K molecules discovered so far. A molecule must have every score required by the objective. Recall is displayed as a percentage.
- Running AUC@K
- Average fractional recall over the paid budget, using left-step integration. Values range from 0 to 1; a higher value means earlier discovery.
- Mean and SD
- Means give equal weight to the 5 displayed targets at the same per-target budget. Sample SD describes target-to-target variation; each target uses seed 100.
- Budget percentage
- 100 × charged measurements / canonical library molecules. DUDE-Z uses the per-target budget.
- Budget reduction
- 100 × (1 − MolScout crossing budget / full budget). An unreached threshold is reported as an outcome, without extrapolation.
AUC@K(B) = sum of fractional Recall@K(b), for b = 0 … B−1, divided by B
REFERENCE