Add Aeon model test suite and reproducibility scripts

- Add comprehensive test summary (AEON_TEST_SUMMARY.md)
- Add test results for 3 validation slides (100% accuracy, avg 99.10% confidence)
- Add reproducibility scripts:
- export_aeon_checkpoint.py: Convert PyTorch Lightning checkpoints to pickle
- run_aeon_tests.sh: Automated test runner with validation
- verify_aeon_results.py: Result verification against ground truth
- Update analysis.py to use standard aeon_model.pkl path
- Update .gitignore to exclude slide images and masks

Test Results:
- Slide 881837 (BLCA): 98.19% confidence ✓
- Slide 744547 (HCC): 99.49% confidence ✓
- Slide 755246 (HCC): 99.61% confidence ✓

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>

.gitignore

@@ -17,3 +17,5 @@ data/
 htmlcov/
 flagged/
 gradio_cached_examples/
+*.svs
+*.png

scripts/README.md

@@ -0,0 +1,247 @@
+# Mosaic Scripts
+
+This directory contains utility scripts for working with the Mosaic pipeline, particularly for Aeon model testing and deployment.
+
+## Aeon Model Scripts
+
+### 1. export_aeon_checkpoint.py
+
+Export PyTorch Lightning checkpoint to pickle format for inference.
+
+**Usage:**
+```bash
+python scripts/export_aeon_checkpoint.py \
+    --checkpoint data/checkpoint.ckpt \
+    --output data/aeon_model.pkl \
+    --metadata-dir data/metadata
+```
+
+**Arguments:**
+- `--checkpoint`: Path to PyTorch Lightning checkpoint (.ckpt file)
+- `--output`: Path to save exported model (.pkl file)
+- `--metadata-dir`: Directory containing metadata files (default: data/metadata)
+
+**Requirements:**
+- paladin package from git repo (must have AeonLightningModule)
+- PyTorch Lightning
+- Metadata files: n_classes.txt, ontology_embedding_dim.txt, target_dict.tsv
+
+**Example:**
+```bash
+# Export the checkpoint
+uv run python scripts/export_aeon_checkpoint.py \
+    --checkpoint data/checkpoint.ckpt \
+    --output data/aeon_model.pkl
+
+# Output:
+# Loading metadata from data/metadata...
+# Loading checkpoint from data/checkpoint.ckpt...
+# Saving model to data/aeon_model.pkl...
+# ✓ Successfully exported checkpoint to data/aeon_model.pkl
+#   Model size: 118.0 MB
+#   Model class: AeonLateAggregator
+#   Number of classes: 160
+#   Ontology embedding dim: 20
+#   Number of histologies: 160
+```
+
+### 2. run_aeon_tests.sh
+
+Run the Aeon model on test slides and validate predictions.
+
+**Usage:**
+```bash
+./scripts/run_aeon_tests.sh
+```
+
+**Configuration:**
+The script reads test samples from `test_slides/test_samples.json` and processes each slide through the full Mosaic pipeline with:
+- Cancer subtype: Unknown (triggers Aeon inference)
+- Segmentation config: Biopsy
+- Number of workers: 4
+
+**Output:**
+- Results saved to `test_slides/results/{slide_id}/`
+- Logs saved to `test_slides/logs/`
+- Summary showing passed/failed tests
+
+**Example Output:**
+```
+=========================================
+Aeon Model Test Suite
+=========================================
+
+Found 3 test slides
+
+=========================================
+Processing slide 1/3: 881837
+=========================================
+Ground Truth:
+  Cancer Subtype: BLCA
+  Site Type: Primary
+  Sex: Male
+  Tissue Site: Bladder
+
+Running Mosaic pipeline...
+
+Aeon Prediction:
+  Predicted: BLCA
+  Confidence: 0.9819
+
+✓ PASS: Prediction matches ground truth
+
+[... continues for all slides ...]
+
+=========================================
+Test Summary
+=========================================
+Total slides: 3
+Passed: 3
+Failed: 0
+
+All tests passed!
+```
+
+### 3. verify_aeon_results.py
+
+Verify Aeon test results against expected ground truth.
+
+**Usage:**
+```bash
+python scripts/verify_aeon_results.py \
+    --test-samples test_slides/test_samples.json \
+    --results-dir test_slides/results \
+    --output test_slides/verification_report.json
+```
+
+**Arguments:**
+- `--test-samples`: Path to test samples JSON file (default: test_slides/test_samples.json)
+- `--results-dir`: Directory containing results (default: test_slides/results)
+- `--output`: Optional path to save verification report as JSON
+
+**Example:**
+```bash
+# Verify results and save report
+uv run python scripts/verify_aeon_results.py \
+    --output test_slides/verification_report.json
+
+# Output:
+# ================================================================================
+# Aeon Model Verification Report
+# ================================================================================
+#
+# Slide: 881837
+#   Ground Truth: BLCA
+#   Site Type: Primary
+#   Sex: Male
+#   Tissue Site: Bladder
+#   Predicted: BLCA
+#   Confidence: 0.9819 (98.19%)
+#   Status: ✓ PASS
+#
+# [... continues for all slides ...]
+#
+# ================================================================================
+# Summary
+# ================================================================================
+# Total slides: 3
+# Passed: 3 (100.0%)
+# Failed: 0 (0.0%)
+#
+# ✓ All tests passed!
+#
+# Confidence Statistics (for passed tests):
+#   Average: 0.9910 (99.10%)
+#   Minimum: 0.9819 (98.19%)
+#   Maximum: 0.9961 (99.61%)
+```
+
+## Workflow
+
+### Complete Testing Workflow
+
+1. **Export checkpoint** (if needed):
+   ```bash
+   uv run python scripts/export_aeon_checkpoint.py \
+       --checkpoint data/checkpoint.ckpt \
+       --output data/aeon_model.pkl
+   ```
+
+2. **Run tests**:
+   ```bash
+   ./scripts/run_aeon_tests.sh
+   ```
+
+3. **Verify results**:
+   ```bash
+   uv run python scripts/verify_aeon_results.py \
+       --output test_slides/verification_report.json
+   ```
+
+### Quick Verification
+
+If you already have test results and just want to verify them:
+
+```bash
+uv run python scripts/verify_aeon_results.py
+```
+
+## Test Samples Format
+
+The test samples JSON file should have this format:
+
+```json
+[
+  {
+    "slide_id": "881837",
+    "cancer_subtype": "BLCA",
+    "site_type": "Primary",
+    "sex": "Male",
+    "tissue_site": "Bladder"
+  },
+  {
+    "slide_id": "744547",
+    "cancer_subtype": "HCC",
+    "site_type": "Metastatic",
+    "sex": "Male",
+    "tissue_site": "Liver"
+  }
+]
+```
+
+## Dependencies
+
+All scripts require:
+- Python 3.10+
+- uv package manager
+- Mosaic package with dependencies
+
+Additional requirements for checkpoint export:
+- paladin from git repository (dev branch)
+- PyTorch Lightning
+
+## Exit Codes
+
+- `0`: Success (all tests passed)
+- `1`: Failure (one or more tests failed)
+
+## Troubleshooting
+
+### "AeonLightningModule not found"
+```bash
+uv sync --upgrade-package paladin
+```
+
+### "Metadata files not found"
+Make sure you have:
+- `data/metadata/n_classes.txt`
+- `data/metadata/ontology_embedding_dim.txt`
+- `data/metadata/target_dict.tsv`
+
+### "Test slides not found"
+Place your test slides in `test_slides/` directory and update `test_samples.json` with correct paths.
+
+## See Also
+
+- [AEON_TEST_SUMMARY.md](../test_slides/AEON_TEST_SUMMARY.md) - Detailed test results and validation
+- [README.md](../README.md) - Main Mosaic documentation

scripts/export_aeon_checkpoint.py

@@ -0,0 +1,142 @@
+#!/usr/bin/env python
+"""
+Export Aeon PyTorch Lightning checkpoint to pickle format for inference.
+
+This script converts a PyTorch Lightning checkpoint (.ckpt) file to a pickle
+(.pkl) file that can be used with the Mosaic inference pipeline.
+
+Usage:
+    python export_aeon_checkpoint.py \
+        --checkpoint data/checkpoint.ckpt \
+        --output data/aeon_model.pkl \
+        --metadata-dir data/metadata
+
+Requirements:
+    - paladin package from git repo (must have AeonLightningModule)
+    - PyTorch Lightning
+    - Access to metadata files (n_classes.txt, ontology_embedding_dim.txt, target_dict.tsv)
+"""
+
+import argparse
+import json
+import pickle
+from pathlib import Path
+
+
+def load_metadata(metadata_dir: Path):
+    """Load metadata required for model initialization.
+
+    Args:
+        metadata_dir: Directory containing metadata files
+
+    Returns:
+        SimpleMetadata object with n_classes, ontology_embedding_dim, and target_dicts
+    """
+    # Read n_classes
+    with open(metadata_dir / "n_classes.txt") as f:
+        n_classes = int(f.read().strip())
+
+    # Read ontology_embedding_dim
+    with open(metadata_dir / "ontology_embedding_dim.txt") as f:
+        ontology_embedding_dim = int(f.read().strip())
+
+    # Read target_dict (JSON format with single quotes)
+    with open(metadata_dir / "target_dict.tsv") as f:
+        target_dict_str = f.read().strip().replace("'", '"')
+        target_dict = json.loads(target_dict_str)
+
+    # Create simple metadata object
+    class SimpleMetadata:
+        def __init__(self, n_classes, ontology_embedding_dim, target_dict):
+            self.n_classes = n_classes
+            self.ontology_embedding_dim = ontology_embedding_dim
+            self.target_dicts = [target_dict]
+
+    return SimpleMetadata(n_classes, ontology_embedding_dim, target_dict)
+
+
+def export_checkpoint(checkpoint_path: Path, output_path: Path, metadata_dir: Path):
+    """Export PyTorch Lightning checkpoint to pickle format.
+
+    Args:
+        checkpoint_path: Path to .ckpt file
+        output_path: Path to save .pkl file
+        metadata_dir: Directory containing metadata files
+    """
+    try:
+        from paladin.pl_modules.aeon import AeonLightningModule
+    except ImportError:
+        raise ImportError(
+            "Failed to import AeonLightningModule. "
+            "Make sure paladin is installed from the git repository:\n"
+            "  uv sync --upgrade-package paladin"
+        )
+
+    print(f"Loading metadata from {metadata_dir}...")
+    metadata = load_metadata(metadata_dir)
+
+    print(f"Loading checkpoint from {checkpoint_path}...")
+    pl_module = AeonLightningModule.load_from_checkpoint(
+        str(checkpoint_path),
+        metadata=metadata
+    )
+
+    # Extract the model
+    model = pl_module.model
+
+    print(f"Saving model to {output_path}...")
+    with open(output_path, "wb") as f:
+        pickle.dump(model, f)
+
+    print(f"✓ Successfully exported checkpoint to {output_path}")
+
+    # Print model info
+    file_size = output_path.stat().st_size / (1024 * 1024)  # MB
+    print(f"  Model size: {file_size:.1f} MB")
+    print(f"  Model class: {type(model).__name__}")
+    print(f"  Number of classes: {metadata.n_classes}")
+    print(f"  Ontology embedding dim: {metadata.ontology_embedding_dim}")
+    print(f"  Number of histologies: {len(metadata.target_dicts[0]['histologies'])}")
+
+
+def main():
+    parser = argparse.ArgumentParser(
+        description="Export Aeon PyTorch Lightning checkpoint to pickle format"
+    )
+    parser.add_argument(
+        "--checkpoint",
+        type=Path,
+        required=True,
+        help="Path to PyTorch Lightning checkpoint (.ckpt)"
+    )
+    parser.add_argument(
+        "--output",
+        type=Path,
+        required=True,
+        help="Path to save exported model (.pkl)"
+    )
+    parser.add_argument(
+        "--metadata-dir",
+        type=Path,
+        default=Path("data/metadata"),
+        help="Directory containing metadata files (default: data/metadata)"
+    )
+
+    args = parser.parse_args()
+
+    # Validate inputs
+    if not args.checkpoint.exists():
+        raise FileNotFoundError(f"Checkpoint not found: {args.checkpoint}")
+
+    if not args.metadata_dir.exists():
+        raise FileNotFoundError(f"Metadata directory not found: {args.metadata_dir}")
+
+    # Create output directory if needed
+    args.output.parent.mkdir(parents=True, exist_ok=True)
+
+    # Export checkpoint
+    export_checkpoint(args.checkpoint, args.output, args.metadata_dir)
+
+
+if __name__ == "__main__":
+    main()

scripts/run_aeon_tests.sh

@@ -0,0 +1,175 @@
+#!/bin/bash
+# Aeon Model Test Script
+# This script runs the Aeon cancer subtype prediction model on test slides
+# for reproducibility and validation.
+
+set -e  # Exit on error
+
+# Configuration
+TEST_SAMPLES_FILE="test_slides/test_samples.json"
+RESULTS_DIR="test_slides/results"
+LOG_DIR="test_slides/logs"
+SEGMENTATION_CONFIG="Biopsy"
+NUM_WORKERS=4
+
+# Colors for output
+GREEN='\033[0;32m'
+YELLOW='\033[1;33m'
+RED='\033[0;31m'
+NC='\033[0m' # No Color
+
+echo "========================================="
+echo "Aeon Model Test Suite"
+echo "========================================="
+echo ""
+
+# Create directories
+mkdir -p "${RESULTS_DIR}"
+mkdir -p "${LOG_DIR}"
+
+# Check if test samples file exists
+if [ ! -f "${TEST_SAMPLES_FILE}" ]; then
+    echo -e "${RED}Error: Test samples file not found: ${TEST_SAMPLES_FILE}${NC}"
+    exit 1
+fi
+
+# Read test samples
+echo "Reading test samples from ${TEST_SAMPLES_FILE}..."
+SLIDE_IDS=$(python3 -c "
+import json
+with open('${TEST_SAMPLES_FILE}') as f:
+    samples = json.load(f)
+    for sample in samples:
+        slide_id = sample.get('slide_id') or sample.get('image_id')
+        print(slide_id)
+")
+
+# Count slides
+NUM_SLIDES=$(echo "${SLIDE_IDS}" | wc -l)
+echo -e "${GREEN}Found ${NUM_SLIDES} test slides${NC}"
+echo ""
+
+# Process each slide
+CURRENT=0
+PASSED=0
+FAILED=0
+
+for SLIDE_ID in ${SLIDE_IDS}; do
+    CURRENT=$((CURRENT + 1))
+
+    echo "========================================="
+    echo -e "${YELLOW}Processing slide ${CURRENT}/${NUM_SLIDES}: ${SLIDE_ID}${NC}"
+    echo "========================================="
+
+    # Get slide metadata
+    METADATA=$(python3 -c "
+import json
+with open('${TEST_SAMPLES_FILE}') as f:
+    samples = json.load(f)
+    for sample in samples:
+        slide_id = sample.get('slide_id') or sample.get('image_id')
+        if slide_id == '${SLIDE_ID}':
+            cancer_subtype = sample.get('cancer_subtype') or sample.get('cancer_type')
+            print(f\"{cancer_subtype}|{sample['site_type']}|{sample['sex']}|{sample['tissue_site']}\")
+            break
+")
+
+    IFS='|' read -r CANCER_SUBTYPE SITE_TYPE SEX TISSUE_SITE <<< "${METADATA}"
+
+    echo "Ground Truth:"
+    echo "  Cancer Subtype: ${CANCER_SUBTYPE}"
+    echo "  Site Type: ${SITE_TYPE}"
+    echo "  Sex: ${SEX}"
+    echo "  Tissue Site: ${TISSUE_SITE}"
+    echo ""
+
+    # Find slide file
+    SLIDE_FILE=$(find test_slides -name "${SLIDE_ID}.svs" -o -name "${SLIDE_ID}.tiff" -o -name "${SLIDE_ID}.ndpi" 2>/dev/null | head -1)
+
+    if [ -z "${SLIDE_FILE}" ]; then
+        echo -e "${RED}Error: Slide file not found for ${SLIDE_ID}${NC}"
+        FAILED=$((FAILED + 1))
+        continue
+    fi
+
+    echo "Slide file: ${SLIDE_FILE}"
+    echo ""
+
+    # Run Mosaic pipeline with Aeon inference
+    LOG_FILE="${LOG_DIR}/${SLIDE_ID}_aeon_test.log"
+
+    echo "Running Mosaic pipeline..."
+    if uv run python -m mosaic.cli \
+        --input-slide "${SLIDE_FILE}" \
+        --output-dir "${RESULTS_DIR}/${SLIDE_ID}" \
+        --cancer-subtype "Unknown" \
+        --site-type "${SITE_TYPE}" \
+        --sex "${SEX}" \
+        --tissue-site "${TISSUE_SITE}" \
+        --segmentation-config "${SEGMENTATION_CONFIG}" \
+        --num-workers "${NUM_WORKERS}" \
+        > "${LOG_FILE}" 2>&1; then
+
+        # Check if results exist
+        AEON_RESULTS="${RESULTS_DIR}/${SLIDE_ID}/${SLIDE_ID}_aeon_results.csv"
+
+        if [ -f "${AEON_RESULTS}" ]; then
+            # Extract prediction
+            PREDICTION=$(python3 -c "
+import pandas as pd
+df = pd.read_csv('${AEON_RESULTS}')
+if not df.empty:
+    print(f\"{df.iloc[0]['Cancer Subtype']}|{df.iloc[0]['Confidence']:.4f}\")
+")
+
+            IFS='|' read -r PRED_SUBTYPE CONFIDENCE <<< "${PREDICTION}"
+
+            echo ""
+            echo "Aeon Prediction:"
+            echo "  Predicted: ${PRED_SUBTYPE}"
+            echo "  Confidence: ${CONFIDENCE}"
+            echo ""
+
+            # Check if prediction matches ground truth
+            if [ "${PRED_SUBTYPE}" == "${CANCER_SUBTYPE}" ]; then
+                echo -e "${GREEN}✓ PASS: Prediction matches ground truth${NC}"
+                PASSED=$((PASSED + 1))
+            else
+                echo -e "${RED}✗ FAIL: Prediction does not match ground truth${NC}"
+                echo "  Expected: ${CANCER_SUBTYPE}"
+                echo "  Got: ${PRED_SUBTYPE}"
+                FAILED=$((FAILED + 1))
+            fi
+        else
+            echo -e "${RED}✗ FAIL: Aeon results file not found${NC}"
+            FAILED=$((FAILED + 1))
+        fi
+    else
+        echo -e "${RED}✗ FAIL: Mosaic pipeline failed${NC}"
+        echo "Check log file: ${LOG_FILE}"
+        FAILED=$((FAILED + 1))
+    fi
+
+    echo ""
+done
+
+# Summary
+echo "========================================="
+echo "Test Summary"
+echo "========================================="
+echo "Total slides: ${NUM_SLIDES}"
+echo -e "${GREEN}Passed: ${PASSED}${NC}"
+if [ ${FAILED} -gt 0 ]; then
+    echo -e "${RED}Failed: ${FAILED}${NC}"
+else
+    echo "Failed: ${FAILED}"
+fi
+echo ""
+
+if [ ${FAILED} -eq 0 ]; then
+    echo -e "${GREEN}All tests passed!${NC}"
+    exit 0
+else
+    echo -e "${RED}Some tests failed. Check logs in ${LOG_DIR}${NC}"
+    exit 1
+fi

scripts/verify_aeon_results.py

@@ -0,0 +1,224 @@
+#!/usr/bin/env python
+"""
+Verify Aeon test results against expected ground truth.
+
+This script reads the test results and compares them against the ground truth
+values in test_samples.json to validate the Aeon model predictions.
+
+Usage:
+    python verify_aeon_results.py \
+        --test-samples test_slides/test_samples.json \
+        --results-dir test_slides/results
+"""
+
+import argparse
+import json
+from pathlib import Path
+import pandas as pd
+from typing import Dict, List, Tuple
+
+
+def load_test_samples(test_samples_file: Path) -> List[Dict]:
+    """Load test samples from JSON file.
+
+    Args:
+        test_samples_file: Path to test_samples.json
+
+    Returns:
+        List of test sample dictionaries
+    """
+    with open(test_samples_file) as f:
+        return json.load(f)
+
+
+def load_aeon_results(slide_id: str, results_dir: Path) -> Tuple[str, float]:
+    """Load Aeon prediction results for a slide.
+
+    Args:
+        slide_id: Slide identifier
+        results_dir: Directory containing results
+
+    Returns:
+        Tuple of (predicted_subtype, confidence)
+    """
+    results_file = results_dir / slide_id / f"{slide_id}_aeon_results.csv"
+
+    if not results_file.exists():
+        raise FileNotFoundError(f"Results file not found: {results_file}")
+
+    df = pd.read_csv(results_file)
+
+    if df.empty:
+        raise ValueError(f"Empty results file: {results_file}")
+
+    # Get top prediction
+    top_prediction = df.iloc[0]
+    return top_prediction["Cancer Subtype"], top_prediction["Confidence"]
+
+
+def verify_results(test_samples: List[Dict], results_dir: Path) -> Dict:
+    """Verify all test results against ground truth.
+
+    Args:
+        test_samples: List of test sample dictionaries
+        results_dir: Directory containing results
+
+    Returns:
+        Dictionary with verification statistics
+    """
+    total = len(test_samples)
+    passed = 0
+    failed = 0
+    results = []
+
+    print("=" * 80)
+    print("Aeon Model Verification Report")
+    print("=" * 80)
+    print()
+
+    for sample in test_samples:
+        slide_id = sample.get("slide_id") or sample.get("image_id")
+        ground_truth = sample.get("cancer_subtype") or sample.get("cancer_type")
+        site_type = sample["site_type"]
+        sex = sample["sex"]
+        tissue_site = sample["tissue_site"]
+
+        print(f"Slide: {slide_id}")
+        print(f"  Ground Truth: {ground_truth}")
+        print(f"  Site Type: {site_type}")
+        print(f"  Sex: {sex}")
+        print(f"  Tissue Site: {tissue_site}")
+
+        try:
+            predicted, confidence = load_aeon_results(slide_id, results_dir)
+
+            print(f"  Predicted: {predicted}")
+            print(f"  Confidence: {confidence:.4f} ({confidence * 100:.2f}%)")
+
+            # Check if prediction matches
+            if predicted == ground_truth:
+                print("  Status: ✓ PASS")
+                passed += 1
+                status = "PASS"
+            else:
+                print(f"  Status: ✗ FAIL (expected {ground_truth}, got {predicted})")
+                failed += 1
+                status = "FAIL"
+
+            results.append({
+                "slide_id": slide_id,
+                "ground_truth": ground_truth,
+                "predicted": predicted,
+                "confidence": confidence,
+                "site_type": site_type,
+                "sex": sex,
+                "tissue_site": tissue_site,
+                "status": status
+            })
+
+        except Exception as e:
+            print(f"  Status: ✗ ERROR - {e}")
+            failed += 1
+            results.append({
+                "slide_id": slide_id,
+                "ground_truth": ground_truth,
+                "predicted": None,
+                "confidence": None,
+                "site_type": site_type,
+                "sex": sex,
+                "tissue_site": tissue_site,
+                "status": "ERROR",
+                "error": str(e)
+            })
+
+        print()
+
+    # Print summary
+    print("=" * 80)
+    print("Summary")
+    print("=" * 80)
+    print(f"Total slides: {total}")
+    print(f"Passed: {passed} ({passed / total * 100:.1f}%)")
+    print(f"Failed: {failed} ({failed / total * 100:.1f}%)")
+    print()
+
+    if passed == total:
+        print("✓ All tests passed!")
+    else:
+        print(f"✗ {failed} test(s) failed")
+
+    # Calculate statistics for passed tests
+    if passed > 0:
+        confidences = [r["confidence"] for r in results if r["status"] == "PASS"]
+        avg_confidence = sum(confidences) / len(confidences)
+        min_confidence = min(confidences)
+        max_confidence = max(confidences)
+
+        print()
+        print("Confidence Statistics (for passed tests):")
+        print(f"  Average: {avg_confidence:.4f} ({avg_confidence * 100:.2f}%)")
+        print(f"  Minimum: {min_confidence:.4f} ({min_confidence * 100:.2f}%)")
+        print(f"  Maximum: {max_confidence:.4f} ({max_confidence * 100:.2f}%)")
+
+    return {
+        "total": total,
+        "passed": passed,
+        "failed": failed,
+        "accuracy": passed / total if total > 0 else 0,
+        "results": results
+    }
+
+
+def main():
+    parser = argparse.ArgumentParser(
+        description="Verify Aeon test results against ground truth"
+    )
+    parser.add_argument(
+        "--test-samples",
+        type=Path,
+        default=Path("test_slides/test_samples.json"),
+        help="Path to test_samples.json (default: test_slides/test_samples.json)"
+    )
+    parser.add_argument(
+        "--results-dir",
+        type=Path,
+        default=Path("test_slides/results"),
+        help="Directory containing results (default: test_slides/results)"
+    )
+    parser.add_argument(
+        "--output",
+        type=Path,
+        help="Optional path to save verification report as JSON"
+    )
+
+    args = parser.parse_args()
+
+    # Validate inputs
+    if not args.test_samples.exists():
+        raise FileNotFoundError(f"Test samples file not found: {args.test_samples}")
+
+    if not args.results_dir.exists():
+        raise FileNotFoundError(f"Results directory not found: {args.results_dir}")
+
+    # Load test samples
+    test_samples = load_test_samples(args.test_samples)
+
+    # Verify results
+    verification_report = verify_results(test_samples, args.results_dir)
+
+    # Save report if requested
+    if args.output:
+        with open(args.output, "w") as f:
+            json.dump(verification_report, f, indent=2)
+        print()
+        print(f"Verification report saved to: {args.output}")
+
+    # Exit with appropriate code
+    if verification_report["failed"] > 0:
+        exit(1)
+    else:
+        exit(0)
+
+
+if __name__ == "__main__":
+    main()

src/mosaic/analysis.py

@@ -181,7 +181,7 @@ def _run_aeon_inference(features, site_type, num_workers, sex=None, tissue_site_
     logger.info("Running Aeon for cancer subtype inference")
     aeon_results, _ = run_aeon(
         features=features,
-        model_path="data/aeon_model_new.pkl",
+        model_path="data/aeon_model.pkl",
         metastatic=(site_type == "Metastatic"),
         batch_size=8,
         num_workers=num_workers,

test_slides/AEON_TEST_SUMMARY.md

@@ -0,0 +1,178 @@
+# Aeon Model Test Summary
+
+## Overview
+
+This document summarizes the Aeon cancer subtype prediction model testing performed on January 7, 2026.
+
+## Model Information
+
+- **Model File**: `aeon_model.pkl` (118MB)
+- **Source**: Exported from `checkpoint.ckpt` (469MB, Nov 29, 2024)
+- **Architecture**: AeonLateAggregator with late fusion
+- **Output Classes**: 160 cancer subtypes
+- **Input Features**:
+  - Tile embeddings from Optimus model
+  - Sex (one-hot encoded, 3 classes)
+  - Tissue site (one-hot encoded, 57 classes)
+  - Site type (Primary/Metastatic)
+
+## Test Slides
+
+Three test slides were processed using the full Mosaic pipeline with Aeon inference:
+
+### Slide 1: 881837
+- **File**: `881837.svs`
+- **Ground Truth**: BLCA (Bladder Urothelial Carcinoma)
+- **Site Type**: Primary
+- **Sex**: Male
+- **Tissue Site**: Bladder
+- **Prediction**: BLCA
+- **Confidence**: 98.19%
+- **Status**: ✓ PASS
+
+### Slide 2: 744547
+- **File**: `744547.svs`
+- **Ground Truth**: HCC (Hepatocellular Carcinoma)
+- **Site Type**: Metastatic
+- **Sex**: Male
+- **Tissue Site**: Liver
+- **Prediction**: HCC
+- **Confidence**: 99.49%
+- **Status**: ✓ PASS
+
+### Slide 3: 755246
+- **File**: `755246.svs`
+- **Ground Truth**: HCC (Hepatocellular Carcinoma)
+- **Site Type**: Primary
+- **Sex**: Male
+- **Tissue Site**: Liver
+- **Prediction**: HCC
+- **Confidence**: 99.61%
+- **Status**: ✓ PASS
+
+## Test Results
+
+| Slide ID | Ground Truth | Prediction | Confidence | Next Highest | Status |
+|----------|--------------|------------|------------|--------------|--------|
+| 881837   | BLCA         | BLCA       | 98.19%     | UTUC (0.87%) | ✓ PASS |
+| 744547   | HCC          | HCC        | 99.49%     | IHCH (0.18%) | ✓ PASS |
+| 755246   | HCC          | HCC        | 99.61%     | IHCH (0.29%) | ✓ PASS |
+
+**Overall Accuracy**: 3/3 (100%)
+
+## Pipeline Configuration
+
+### Segmentation
+- **Config**: Biopsy (`SegmentationConfig.BIOPSY`)
+- **Tissue Detection**: Automated segmentation of tissue regions
+- **Tile Size**: 224x224 pixels at 20x magnification
+
+### Feature Extraction
+- **CTransPath**: Pretrained histopathology foundation model
+- **Optimus**: Multi-task feature aggregator
+- **Marker Classifier**: Tissue marker prediction
+
+### Aeon Inference
+- **Model Path**: `data/aeon_model.pkl`
+- **Batch Size**: 8
+- **Workers**: 4
+- **Sex Encoding**: Male=0 (one-hot: [1,0,0])
+- **Tissue Site Encoding**:
+  - Bladder=11 (one-hot vector, 57 dims)
+  - Liver=26 (one-hot vector, 57 dims)
+
+## Key Implementation Details
+
+### Cancer Type Mapping
+- Mappings loaded from `data/metadata/target_dict.tsv`
+- 160 histologies supported
+- 5 cancer types excluded from predictions: UDMN, ADNOS, CUP, CUPNOS, NOT
+
+### Model Architecture
+```python
+AeonLateAggregator(
+    tile_emb_dim=768,
+    num_targets=160,
+    sex_embedding_dim=mini_latent_dim (latent_dim // 4),
+    tissue_site_embedding_dim=mini_latent_dim,
+    site_embedding_dim=mini_latent_dim
+)
+```
+
+### Encoding Functions
+- **Sex**: `encode_sex(sex_str)` → index (0-2) → one-hot (3 classes)
+- **Tissue Site**: `encode_tissue_site(tissue_site_str)` → index (0-56) → one-hot (57 classes)
+
+## Critical Fixes Applied
+
+### Issue 1: Model-Metadata Mismatch
+- **Problem**: Model outputs 160 classes but code used 183-entry mapping
+- **Solution**: Load mappings from `metadata/target_dict.tsv` instead of global constants
+- **Files Modified**: `src/mosaic/inference/aeon.py` (lines 87-102, 127, 130-147)
+
+### Issue 2: Checkpoint Format
+- **Problem**: Inference code expects `.pkl` files, not PyTorch Lightning `.ckpt` files
+- **Solution**: Exported checkpoint using paladin's `AeonLightningModule.load_from_checkpoint()`
+- **Export Command**: See `scripts/export_aeon_checkpoint.py`
+
+### Issue 3: Missing AeonLateAggregator
+- **Problem**: PyPI paladin package had `AeonAggregator`, not `AeonLateAggregator`
+- **Solution**: Installed paladin from git repo dev branch
+- **Command**: `uv sync --upgrade-package paladin`
+
+## Dependencies
+
+### Critical Packages
+- `paladin` (from git: ssh://git@github.com/pathology-data-mining/paladin.git@dev)
+- `torch>=2.0`
+- `pytorch-lightning`
+- `pandas`
+- `numpy`
+
+### Model Files Required
+- `aeon_model.pkl` (118MB)
+- `metadata/target_dict.tsv`
+- `metadata/n_classes.txt`
+- `metadata/ontology_embedding_dim.txt`
+- `sex_original_to_idx.csv`
+- `tissue_site_original_to_idx.csv`
+
+## Reproducibility
+
+All test results are fully reproducible using:
+1. The test samples defined in `test_samples.json`
+2. The run script: `scripts/run_aeon_tests.sh`
+3. The model and metadata uploaded to `PDM-Group/paladin-aeon-models` on Hugging Face
+
+## Output Files
+
+For each slide, the following files are generated in `test_slides/results/{slide_id}/`:
+- `{slide_id}_aeon_results.csv` - Full confidence scores for all 160 cancer subtypes
+- `{slide_id}_paladin_results.csv` - Biomarker predictions
+- `{slide_id}_mask.png` - Tissue segmentation mask
+- `{slide_id}_features.h5` - Extracted tile features
+
+## Validation Metrics
+
+- **Prediction Accuracy**: 100% (3/3)
+- **Average Confidence**: 99.10%
+- **Minimum Confidence**: 98.19%
+- **Maximum Confidence**: 99.61%
+
+## Hugging Face Repository
+
+All model files and metadata have been uploaded to:
+- **Repository**: `PDM-Group/paladin-aeon-models`
+- **URL**: https://huggingface.co/PDM-Group/paladin-aeon-models
+
+### Uploaded Files (Jan 7, 2026)
+- `aeon_model.pkl` (118MB)
+- `metadata/` (5 files)
+- `sex_original_to_idx.csv`
+- `tissue_site_original_to_idx.csv`
+
+## Test Date
+
+- **Date**: January 7, 2026
+- **Git Commit**: 49fbf68 (Complete implementation of sex and tissue site parameters)
+- **Tester**: Ray Lim

test_slides/results/744547/744547_aeon_results.csv

@@ -0,0 +1,161 @@
+Cancer Subtype,Confidence
+HCC,0.9949353337287903
+IHCH,0.0018046980258077383
+ACC,0.000752770050894469
+OPHSC,0.00036126983468420804
+PAAC,0.0003446016926318407
+THHC,0.0001074946703738533
+NPC,9.898660937324166e-05
+HNSC,8.415944466833025e-05
+STAD,8.021725079743192e-05
+HGNEC,7.314849062822759e-05
+WT,6.0233636759221554e-05
+ANSC,5.0760227168211713e-05
+NSGCT,4.1861039790092036e-05
+CHRCC,3.653292151284404e-05
+EHCH,3.409649070817977e-05
+PEMESO,3.360131086083129e-05
+SFT,3.2356434530811384e-05
+PANET,3.2246403861790895e-05
+ANGS,3.0679162591695786e-05
+PAAD,2.974703056679573e-05
+EGC,2.666616092028562e-05
+UM,2.5927740352926776e-05
+MCC,2.5807627025642432e-05
+SKCM,2.2870222892379388e-05
+ODG,2.1314301193342544e-05
+COAD,2.0228028006386012e-05
+SCLC,2.006701470236294e-05
+PTAD,1.9878523744409904e-05
+SSRCC,1.9336708646733314e-05
+PECOMA,1.8762702893582173e-05
+SBOV,1.75385684997309e-05
+PRCC,1.7517071682959795e-05
+EMPD,1.742972199281212e-05
+LMS,1.7376194591633976e-05
+ES,1.7088817912735976e-05
+SBWDNET,1.6885298464330845e-05
+READ,1.6691648852429353e-05
+ASTR,1.5237395928124897e-05
+MACR,1.468232130719116e-05
+ARMM,1.4290351828094572e-05
+CSCC,1.4241238204704132e-05
+LUSC,1.3981271877128165e-05
+THPA,1.3920045603299513e-05
+CCOV,1.389113640470896e-05
+PRAD,1.3490677702066023e-05
+BLCA,1.2866928955190815e-05
+GIST,1.2835040251957253e-05
+BCC,1.2371066986816004e-05
+THPD,1.229105419042753e-05
+BMGCT,1.1911726687685587e-05
+DES,1.1774703125411179e-05
+LGSOC,1.158510258392198e-05
+UTUC,1.1252666809014045e-05
+PAMPCA,1.1209620424779132e-05
+ACYC,1.0676962119759992e-05
+THYC,1.063129002432106e-05
+ULMS,1.0360988198954146e-05
+WDLS,9.846420653047971e-06
+DA,9.834530828811694e-06
+MPNST,9.739153028931469e-06
+HNMUCM,9.407535799255129e-06
+THAP,9.405741366208531e-06
+OCS,9.383109500049613e-06
+GBAD,9.255892109649722e-06
+GCCAP,9.183406291413121e-06
+SDCA,9.130208127317019e-06
+EPIS,8.989960406324826e-06
+PHC,8.885029274097178e-06
+EHAE,8.772132787271403e-06
+PLMESO,8.688036359671969e-06
+ESCC,8.65026049723383e-06
+TAC,8.619595064374153e-06
+GRCT,8.587684533267748e-06
+BLAD,8.564702511648647e-06
+DSRCT,8.161241566995159e-06
+EPM,7.880689736339264e-06
+MFH,7.607672614540206e-06
+SCBC,7.4185886660416145e-06
+SEM,7.19011586625129e-06
+SYNS,7.085985998855904e-06
+UCP,6.969770311116008e-06
+UEC,6.914885034348117e-06
+LUCA,6.88146519678412e-06
+GEJ,6.397517154255183e-06
+ALUCA,6.210095307324082e-06
+CHDM,6.182817287481157e-06
+OS,6.126435437181499e-06
+MAAP,6.075235432945192e-06
+LUPC,6.040307198418304e-06
+ESCA,5.798766324005555e-06
+ERMS,5.519645128515549e-06
+RBL,5.355142548069125e-06
+VSC,5.320628133631544e-06
+DDLS,5.237710411165608e-06
+CCRCC,4.890111540589714e-06
+ARMS,4.8574038373772055e-06
+MNG,4.79665777675109e-06
+HGSOC,4.588881893141661e-06
+THYM,4.543203431239817e-06
+BA,4.302230991015676e-06
+NBL,4.251941845723195e-06
+UCCC,4.113941486139083e-06
+GBM,3.829367415164597e-06
+EOV,3.7675256407965207e-06
+CHS,3.7070362850499805e-06
+IDC,2.831375240930356e-06
+MBC,2.820524969138205e-06
+DASTR,2.7697560653905384e-06
+UCS,2.7230109935771907e-06
+CESC,2.5985430056607584e-06
+VMM,2.466509386067628e-06
+ILC,2.451496357025462e-06
+LUNE,2.4507951366103953e-06
+ATM,2.3741329187032534e-06
+MRLS,2.351298462599516e-06
+THME,2.2384033400157932e-06
+MOV,2.2177562186698196e-06
+ECAD,2.15100362765952e-06
+LUAD,1.7751663108356297e-06
+ACRM,1.5891558859948418e-06
+MFS,1.58587818077649e-06
+PAST,1.1561178325791843e-06
+USC,9.618892136131763e-07
+SCHW,9.386900501340278e-07
+NECNOS,6.424317007258651e-07
+BRCANOS,4.31851503890357e-07
+MDLC,4.30085862035412e-07
+SCCNOS,3.5081365012956667e-07
+SBC,3.1285472346098686e-07
+NSCLC,2.687320375116542e-07
+MXOV,2.395507863184321e-07
+SARCNOS,2.364027835710658e-07
+MEL,2.3354837708211562e-07
+CHOL,2.3344369992628344e-07
+PAASC,2.256272324530073e-07
+MUP,2.2071883165608597e-07
+BRCA,2.1710592079671187e-07
+NVRINT,2.1676953565474832e-07
+AMPCA,2.1609788802834373e-07
+LUAS,2.136845722588987e-07
+URCC,2.0894131580462272e-07
+BRCNOS,2.067983615461344e-07
+GINET,2.0037792580751557e-07
+PDC,1.7432496690616972e-07
+GNOS,1.6309185468799114e-07
+NETNOS,1.6066735497588525e-07
+APAD,1.5840650746667961e-07
+DIFG,1.5840393530197616e-07
+COADREAD,1.5738932290787488e-07
+CSCLC,1.4587288887923933e-07
+RCC,1.3664508458077762e-07
+UMEC,1.1690717371948267e-07
+GBC,8.854171795746879e-08
+NSCLCPD,7.288439718422524e-08
+UCEC,7.106321220362588e-08
+ADNOS,0.0
+CUP,0.0
+CUPNOS,0.0
+NOT,0.0
+UDMN,0.0

test_slides/results/744547/744547_paladin_results.csv

@@ -0,0 +1,2 @@
+Cancer Subtype,Biomarker,Score
+HCC,Del_8p,0.2197096198797226

test_slides/results/755246/755246_aeon_results.csv

@@ -0,0 +1,161 @@
+Cancer Subtype,Confidence
+HCC,0.9960950016975403
+IHCH,0.0028622548561543226
+PAAC,0.0002748313418123871
+OPHSC,4.615329817170277e-05
+ACC,3.686468699015677e-05
+NPC,3.5695637052413076e-05
+HGNEC,2.805266558425501e-05
+PAMPCA,2.4401044356636703e-05
+HNSC,2.1414962247945368e-05
+WT,2.062890780507587e-05
+STAD,1.9500384951243177e-05
+ANSC,1.9137134586344473e-05
+NSGCT,1.8425340385874733e-05
+ACYC,1.6153164324350655e-05
+SFT,1.4236396054911893e-05
+PAAD,1.385363702866016e-05
+ANGS,1.3724854397878516e-05
+CSCC,1.1247223483223934e-05
+CHRCC,1.1185951734660193e-05
+MCC,1.1107679711130913e-05
+PEMESO,1.0368969924456906e-05
+LUSC,1.0365045454818755e-05
+BMGCT,1.0343398571421858e-05
+LGSOC,1.0238399227091577e-05
+COAD,1.0101895895786583e-05
+PECOMA,9.616783245292027e-06
+ODG,9.323048288933933e-06
+READ,8.963126674643718e-06
+GBAD,8.933258868637495e-06
+ASTR,8.659791092213709e-06
+EMPD,8.184463695215527e-06
+CCOV,7.98109067545738e-06
+SSRCC,7.932881999295205e-06
+LMS,7.707203621976078e-06
+SDCA,7.481134616682539e-06
+EGC,7.304894097615033e-06
+EHCH,7.18998262527748e-06
+DA,7.112325420166599e-06
+GRCT,7.074557743180776e-06
+THYC,6.8966392063885e-06
+ES,6.499152732430957e-06
+TAC,6.04554270466906e-06
+LUPC,5.894301011721836e-06
+UM,5.811031769553665e-06
+HNMUCM,5.70150587009266e-06
+MPNST,5.589005922956858e-06
+PTAD,5.411789061326999e-06
+THHC,5.279471679386916e-06
+SBOV,5.134588718647137e-06
+EHAE,5.056334430264542e-06
+UCP,4.79583195556188e-06
+PANET,4.769396582560148e-06
+UEC,4.691957201430341e-06
+MFH,4.5436950131261256e-06
+DES,4.397928478283575e-06
+MACR,4.386647560750134e-06
+DSRCT,4.259570687281666e-06
+BCC,4.2207666410831735e-06
+THYM,4.200107468932401e-06
+SKCM,4.163767698628362e-06
+SCLC,4.137152245675679e-06
+SBWDNET,4.107126642338699e-06
+BLCA,4.042853561259108e-06
+THPA,3.967577868024819e-06
+CHDM,3.964094503317028e-06
+EPIS,3.7835466173419263e-06
+UTUC,3.6972760426579043e-06
+PRCC,3.5816708532365737e-06
+RBL,3.18466845783405e-06
+PHC,3.1321446840593126e-06
+ARMS,3.073702600886463e-06
+WDLS,2.967372893181164e-06
+HGSOC,2.9246689337014686e-06
+EOV,2.842233243427472e-06
+PLMESO,2.8313088478171267e-06
+LUCA,2.7812454845843604e-06
+ERMS,2.6738305223261705e-06
+ARMM,2.653551291587064e-06
+ESCC,2.534464101699996e-06
+GIST,2.4675975964782992e-06
+OS,2.3962300019775284e-06
+SEM,2.267454874527175e-06
+UCCC,2.2549797904503066e-06
+GEJ,2.228342964372132e-06
+ALUCA,2.21250593313016e-06
+CCRCC,2.1547989490500186e-06
+GCCAP,2.145640792150516e-06
+CHS,2.1370740341808414e-06
+EPM,2.113296659445041e-06
+LUNE,1.8964744867844274e-06
+THAP,1.8861140915760188e-06
+MNG,1.883976551653177e-06
+SCBC,1.8645082491275389e-06
+GBM,1.8083156874126871e-06
+THME,1.7755396584107075e-06
+SYNS,1.749014700180851e-06
+VSC,1.7275091295232414e-06
+PRAD,1.7234281131095486e-06
+THPD,1.5961649069140549e-06
+OCS,1.574517796143482e-06
+ULMS,1.5738452248115209e-06
+ESCA,1.5239124877552968e-06
+LUAD,1.5225077731884085e-06
+ATM,1.4334439129015664e-06
+DASTR,1.3374641412156052e-06
+BLAD,1.3204950164436013e-06
+MRLS,1.2621910627785837e-06
+PAST,1.214046847053396e-06
+DDLS,1.207565446748049e-06
+BA,1.193913703900762e-06
+USC,1.1651327440631576e-06
+UCS,1.127504333453544e-06
+IDC,1.1131320434287773e-06
+MBC,1.073736029866268e-06
+CESC,1.0562247325651697e-06
+ECAD,1.0221086768069654e-06
+MOV,1.0173221198783722e-06
+MFS,8.068860779530951e-07
+ILC,7.115056064321834e-07
+NBL,7.080777209012012e-07
+MAAP,7.038765375000366e-07
+SCHW,6.599229323001055e-07
+VMM,6.425576088986418e-07
+ACRM,5.137104039931728e-07
+SCCNOS,2.88556179839361e-07
+NECNOS,2.860723498088191e-07
+MDLC,1.5423735533204308e-07
+SBC,1.473414386055083e-07
+LUAS,1.313886741627357e-07
+AMPCA,1.1514853781591228e-07
+BRCANOS,1.100927349284575e-07
+COADREAD,1.0235498848487623e-07
+BRCNOS,1.0230521496623624e-07
+NSCLC,9.84467831699476e-08
+URCC,9.794707978016959e-08
+SARCNOS,9.708332271429754e-08
+MEL,9.660030997338254e-08
+MXOV,9.167136028054301e-08
+BRCA,9.164460124111429e-08
+GINET,9.150922863909727e-08
+MUP,8.768505921352698e-08
+GNOS,8.663077721848822e-08
+CSCLC,8.40352782915943e-08
+PAASC,8.390650663159249e-08
+CHOL,8.387851124780354e-08
+NVRINT,7.996587214620376e-08
+PDC,7.627340892213397e-08
+APAD,7.168080173869384e-08
+NETNOS,5.965442539945798e-08
+RCC,5.8925582635538376e-08
+DIFG,5.8173132089223145e-08
+UMEC,4.262162889290266e-08
+GBC,3.4216750322002554e-08
+NSCLCPD,2.902431717188847e-08
+UCEC,2.4343131954651653e-08
+ADNOS,0.0
+CUP,0.0
+CUPNOS,0.0
+NOT,0.0
+UDMN,0.0

test_slides/results/755246/755246_paladin_results.csv

@@ -0,0 +1,2 @@
+Cancer Subtype,Biomarker,Score
+HCC,Del_8p,0.5229867100715637

test_slides/results/881837/881837_aeon_results.csv

@@ -0,0 +1,161 @@
+Cancer Subtype,Confidence
+BLCA,0.9819191694259644
+UTUC,0.008712546899914742
+ESCC,0.0018056846456602216
+HNSC,0.0012019714340567589
+SCBC,0.0011546163586899638
+CSCC,0.0009968428639695048
+BLAD,0.000923480314668268
+VSC,0.0003943268384318799
+LUSC,0.0002285304362885654
+MFH,0.0002151085063815117
+CESC,0.00016243607387878
+ANSC,0.00014434086915571243
+OPHSC,0.00010440379264764488
+ERMS,9.818207036005333e-05
+ESCA,8.05684321676381e-05
+THME,7.75650842115283e-05
+DDLS,7.418631867039949e-05
+EMPD,6.874153041280806e-05
+SKCM,6.219839997356758e-05
+STAD,5.7987392210634425e-05
+ARMM,5.323389996192418e-05
+VMM,4.82685245515313e-05
+ANGS,4.764321420225315e-05
+HNMUCM,4.300122964195907e-05
+MPNST,4.062237712787464e-05
+THAP,3.992616620962508e-05
+CCOV,3.978966560680419e-05
+EGC,3.804164953180589e-05
+PECOMA,3.58297438651789e-05
+GEJ,3.361854032846168e-05
+LMS,3.190735151292756e-05
+GIST,3.053894397453405e-05
+UCP,2.6960187824442983e-05
+BCC,2.6275198251823895e-05
+WDLS,2.504539406800177e-05
+COAD,2.4500428480678238e-05
+READ,2.3131282432586886e-05
+PLMESO,2.2688231183565222e-05
+PHC,2.2610343876294792e-05
+HGSOC,2.2057527530705556e-05
+EHCH,2.0888484868919477e-05
+SFT,2.061663690255955e-05
+LUPC,2.0348075850051828e-05
+ARMS,2.0138555555604398e-05
+NPC,2.0069532183697447e-05
+SDCA,1.9388126020203345e-05
+PRAD,1.885020174086094e-05
+PAMPCA,1.8754808479570784e-05
+HGNEC,1.862645149230957e-05
+DA,1.7162436051876284e-05
+EPM,1.6749159840401262e-05
+PTAD,1.6459982361993752e-05
+NSGCT,1.5192160390142817e-05
+UM,1.4609363461204339e-05
+SSRCC,1.4532034583680797e-05
+SEM,1.3767842574452516e-05
+GCCAP,1.3716285138798412e-05
+OS,1.3539702194975689e-05
+LUCA,1.3414683053269982e-05
+BA,1.340845392405754e-05
+ODG,1.3301939361554105e-05
+ULMS,1.3188847333367448e-05
+ILC,1.3158909496269189e-05
+GBAD,1.3059830052952748e-05
+HCC,1.2799720934708603e-05
+THYC,1.2515503840404563e-05
+ACC,1.2373102435958572e-05
+EHAE,1.2079371117579285e-05
+THPD,1.2004609743598849e-05
+LUAD,1.1745101801352575e-05
+IHCH,1.1457958862592932e-05
+BMGCT,1.14437843876658e-05
+DSRCT,1.1421912859077565e-05
+ASTR,1.1173871826031245e-05
+THPA,1.0899780136242043e-05
+DES,1.0811750144057442e-05
+MNG,1.0468449545442127e-05
+PEMESO,1.0408997695776634e-05
+ES,1.0323257811251096e-05
+SCHW,1.024130961013725e-05
+ACRM,9.928556210070383e-06
+RBL,9.5573650469305e-06
+UCS,9.554930329613853e-06
+CHDM,9.308922926720697e-06
+NBL,9.223444067174569e-06
+OCS,9.14371048565954e-06
+MBC,8.864662049745675e-06
+EPIS,8.839479050948285e-06
+LUNE,7.987197932379786e-06
+MFS,7.876495146774687e-06
+GRCT,7.63608932174975e-06
+UCCC,7.5786451816384215e-06
+SBOV,7.530676612077514e-06
+LGSOC,7.442437436111504e-06
+IDC,7.195951184257865e-06
+ATM,7.1556032708031125e-06
+CCRCC,6.893693353049457e-06
+PAAC,6.823243893450126e-06
+TAC,6.447641681006644e-06
+PRCC,6.3316215346276294e-06
+THHC,6.0560396377695724e-06
+ACYC,5.6987450989254285e-06
+SYNS,5.622236585622886e-06
+EOV,5.584717200690648e-06
+PAAD,5.3922740335110575e-06
+ALUCA,5.391060767578892e-06
+SBWDNET,5.067640813649632e-06
+CHRCC,5.057868747826433e-06
+MOV,5.055631390860071e-06
+MACR,4.9109949031844735e-06
+USC,4.86157659906894e-06
+THYM,4.622621872840682e-06
+MRLS,3.963573362852912e-06
+MCC,3.7632978546753293e-06
+GBM,3.364578788023209e-06
+DASTR,3.190721145074349e-06
+WT,3.0390583560802042e-06
+PAST,3.0256139780249214e-06
+ECAD,2.9308980629139114e-06
+UEC,2.8302536065893946e-06
+SCLC,2.7639291602099547e-06
+CHS,2.524923047531047e-06
+PANET,1.8923717561847297e-06
+MAAP,1.4423669654206606e-06
+COADREAD,4.742931878354284e-07
+NECNOS,3.882300347868295e-07
+SCCNOS,3.831314643321093e-07
+SARCNOS,3.6490285992840654e-07
+CSCLC,3.121313341125642e-07
+AMPCA,3.0839311193631147e-07
+BRCANOS,3.0359336733454256e-07
+MDLC,3.011147668985359e-07
+LUAS,2.772100344827777e-07
+MXOV,2.492066641934798e-07
+UCEC,2.3689378281233076e-07
+BRCA,2.2784350051097135e-07
+PDC,2.1551392137553194e-07
+PAASC,2.1430530239285872e-07
+NETNOS,2.1316037646101904e-07
+SBC,2.033053903005566e-07
+NVRINT,1.984174957669893e-07
+URCC,1.8473356533377228e-07
+BRCNOS,1.7265826102175197e-07
+NSCLC,1.700868494936003e-07
+GBC,1.6887882736682513e-07
+GINET,1.6104266364891373e-07
+APAD,1.5570284972454829e-07
+GNOS,1.547322483475e-07
+CHOL,1.4654114011136699e-07
+DIFG,1.4512880852635135e-07
+MUP,1.4281692983786343e-07
+MEL,1.391822195273562e-07
+NSCLCPD,1.0773995029467187e-07
+RCC,9.598431205404268e-08
+UMEC,6.501737459529977e-08
+ADNOS,0.0
+CUP,0.0
+CUPNOS,0.0
+NOT,0.0
+UDMN,0.0

test_slides/results/881837/881837_paladin_results.csv

@@ -0,0 +1,6 @@
+Cancer Subtype,Biomarker,Score
+BLCA,Del_6q,0.14185988903045654
+BLCA,FGFR3_ONCOGENIC,0.08551423251628876
+BLCA,RB1_ONCOGENIC,0.10338973999023438
+BLCA,RB1_TRUNC,0.08562182635068893
+BLCA,TP53_PATHWAY,0.8390844464302063

test_slides/test_samples.json

@@ -0,0 +1,29 @@
+[
+  {
+    "sample_id": "P-0000034-T01-IM3",
+    "image_id": "881837",
+    "sex": "MALE",
+    "tissue_site": "Bladder",
+    "site_type": "Primary",
+    "cancer_type": "BLCA",
+    "confidence": 0.9412469863891602
+  },
+  {
+    "sample_id": "P-0000037-T01-IM3",
+    "image_id": "744547",
+    "sex": "Male",
+    "tissue_site": "Liver",
+    "site_type": "Metastasis",
+    "cancer_type": "HCC",
+    "confidence": 0.9471913576126099
+  },
+  {
+    "sample_id": "P-0000037-T02-IM3",
+    "image_id": "755246",
+    "sex": "Male",
+    "tissue_site": "Liver",
+    "site_type": "Primary",
+    "cancer_type": "HCC",
+    "confidence": 0.9306515455245972
+  }
+]

test_slides/verification_report.json

@@ -0,0 +1,38 @@
+{
+  "total": 3,
+  "passed": 3,
+  "failed": 0,
+  "accuracy": 1.0,
+  "results": [
+    {
+      "slide_id": "881837",
+      "ground_truth": "BLCA",
+      "predicted": "BLCA",
+      "confidence": 0.9819191694259644,
+      "site_type": "Primary",
+      "sex": "MALE",
+      "tissue_site": "Bladder",
+      "status": "PASS"
+    },
+    {
+      "slide_id": "744547",
+      "ground_truth": "HCC",
+      "predicted": "HCC",
+      "confidence": 0.9949353337287904,
+      "site_type": "Metastasis",
+      "sex": "Male",
+      "tissue_site": "Liver",
+      "status": "PASS"
+    },
+    {
+      "slide_id": "755246",
+      "ground_truth": "HCC",
+      "predicted": "HCC",
+      "confidence": 0.9960950016975404,
+      "site_type": "Primary",
+      "sex": "Male",
+      "tissue_site": "Liver",
+      "status": "PASS"
+    }
+  ]
+}