Spaces:
Runtime error
Runtime error
Update README.md
Browse files
README.md
CHANGED
|
@@ -4,7 +4,7 @@ emoji: 🧬
|
|
| 4 |
colorFrom: blue
|
| 5 |
colorTo: indigo
|
| 6 |
sdk: gradio
|
| 7 |
-
sdk_version:
|
| 8 |
python_version: "3.11"
|
| 9 |
app_file: app.py
|
| 10 |
pinned: false
|
|
@@ -33,8 +33,6 @@ short_description: Explanation-first tri-model genomic variant analysis
|
|
| 33 |
|
| 34 |
## Internal signals extracted
|
| 35 |
|
| 36 |
-
From each model — nothing redesigned, internal hooks only:
|
| 37 |
-
|
| 38 |
- **conv3 activation norm profile** — per-nucleotide CNN signal (99 positions)
|
| 39 |
- **Gradient attribution map** — input-gradient backward pass
|
| 40 |
- **Mutation-centered peak** — activation value at the specific mutation position
|
|
@@ -43,17 +41,6 @@ From each model — nothing redesigned, internal hooks only:
|
|
| 43 |
- **Feature ablation response** — causal effect of zeroing splice / region / mutation features
|
| 44 |
- **Risk tier classification** — PATHOGENIC → BENIGN
|
| 45 |
|
| 46 |
-
## Cross-model explainability metrics
|
| 47 |
-
|
| 48 |
-
| Metric | Description |
|
| 49 |
-
|---|---|
|
| 50 |
-
| Mutation Peak Ratio | peak_signal / mean_signal across window |
|
| 51 |
-
| Counterfactual Magnitude | max probability range across alternatives |
|
| 52 |
-
| Cross-Model Locality Score | cosine similarity of conv3 profiles |
|
| 53 |
-
| Signal Concentration Index | fraction of activation within ±5bp of mutation |
|
| 54 |
-
| Explainability Strength Score | composite 0–1 score |
|
| 55 |
-
| Activation Pattern Type | Sharp / Broad / Flat |
|
| 56 |
-
|
| 57 |
## ⚠️ Disclaimer
|
| 58 |
|
| 59 |
For research use only. Not a clinical diagnostic tool.
|
|
|
|
| 4 |
colorFrom: blue
|
| 5 |
colorTo: indigo
|
| 6 |
sdk: gradio
|
| 7 |
+
sdk_version: 4.44.1
|
| 8 |
python_version: "3.11"
|
| 9 |
app_file: app.py
|
| 10 |
pinned: false
|
|
|
|
| 33 |
|
| 34 |
## Internal signals extracted
|
| 35 |
|
|
|
|
|
|
|
| 36 |
- **conv3 activation norm profile** — per-nucleotide CNN signal (99 positions)
|
| 37 |
- **Gradient attribution map** — input-gradient backward pass
|
| 38 |
- **Mutation-centered peak** — activation value at the specific mutation position
|
|
|
|
| 41 |
- **Feature ablation response** — causal effect of zeroing splice / region / mutation features
|
| 42 |
- **Risk tier classification** — PATHOGENIC → BENIGN
|
| 43 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 44 |
## ⚠️ Disclaimer
|
| 45 |
|
| 46 |
For research use only. Not a clinical diagnostic tool.
|