Spaces:
Running
Running
Miyu Horiuchi
Final demo deliverable: v1 baseline + 5K uncultured predictions + recommender online
79a721f | Inputs: 17,047 feature rows, 38,649 strain↔medium links | |
| Training table: 12,689 strains × 353 features × 24 media | |
| Distinct families: 562 | |
| Trained 24 per-medium classifiers in 302.3s | |
| Wrote /Users/miyuhoriuchi/microbe-model/artifacts/media_recommender_results.json | |
| Fitting production models on full dataset... | |
| Saved 24 production models to /Users/miyuhoriuchi/microbe-model/models/recommender | |
| Median PR-AUC: 0.180 | |
| Median ROC-AUC: 0.860 | |
| Top 15 best-modeled media (by PR-AUC): | |
| medium_id name n_pos n_neg pr_auc roc_auc | |
| 65 GYM STREPTOMYCES MEDIUM 1608 11081 0.730911 0.963128 | |
| 514 BACTO MARINE BROTH (DIFCO 2216) 1319 11370 0.574440 0.910297 | |
| 84 ROLLED OATS MINERAL MEDIUM 599 12090 0.370680 0.953196 | |
| 830 R2A MEDIUM 1202 11487 0.325919 0.839716 | |
| 693 COLUMBIA BLOOD MEDIUM 1415 11274 0.323821 0.793217 | |
| 92 TRYPTICASE SOY YEAST EXTRACT MEDIUM 1215 11474 0.304993 0.790605 | |
| 1 NUTRIENT AGAR 1045 11644 0.297507 0.865132 | |
| 104 PYG MEDIUM (modified) 349 12340 0.232898 0.904692 | |
| 553 GPHF-MEDIUM 276 12413 0.211961 0.901489 | |
| 220 CASO AGAR (Merck 105458) 491 12198 0.189578 0.855436 | |
| 339 WILKINS-CHALGREN ANAEROBE BROTH 123 12566 0.189173 0.881104 | |
| 78 CHOPPED MEAT MEDIUM 185 12504 0.186303 0.937636 | |
| 535 TRYPTICASE SOY BROTH AGAR 996 11693 0.173008 0.708869 | |
| 110 CHOPPED MEAT MEDIUM WITH CARBOHYDRATES 164 12525 0.150335 0.897556 | |
| 554 N-Z-AMINE-MEDIUM 209 12480 0.128135 0.880503 | |
| Worst 5: | |
| medium_id name n_pos n_neg pr_auc roc_auc | |
| 215 BHI MEDIUM 257 12432 0.040803 0.589778 | |
| 545 TRYPTONE SOYA BROTH (TSB) 154 12535 0.031110 0.705497 | |
| 381 LB (Luria-Bertani) MEDIUM 130 12559 0.018043 0.638446 | |
| 58 BIFIDOBACTERIUM MEDIUM 129 12560 0.017828 0.709778 | |
| 645 MIDDLEBROOK MEDIUM 202 12487 0.005997 0.763584 | |