Spaces:
Sleeping
Sleeping
Upload app.py with huggingface_hub
Browse files
app.py
CHANGED
|
@@ -7,8 +7,33 @@ from datasets import load_dataset
|
|
| 7 |
|
| 8 |
# Load the dataset
|
| 9 |
print("Loading dataset...")
|
| 10 |
-
|
| 11 |
-
print(f"Loaded {len(
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 12 |
|
| 13 |
# Colors for highlighting
|
| 14 |
COLORS = [
|
|
@@ -60,16 +85,13 @@ def highlight_text(cr_text, annotation):
|
|
| 60 |
color_idx = 0
|
| 61 |
|
| 62 |
for s in filtered:
|
| 63 |
-
# Text before span
|
| 64 |
if s["start"] > pos:
|
| 65 |
html.append(escape_html(cr_text[pos:s["start"]]))
|
| 66 |
|
| 67 |
-
# Assign color
|
| 68 |
if s["var"] not in color_map:
|
| 69 |
color_map[s["var"]] = COLORS[color_idx % len(COLORS)]
|
| 70 |
color_idx += 1
|
| 71 |
|
| 72 |
-
# Highlighted span
|
| 73 |
color = color_map[s["var"]]
|
| 74 |
html.append(
|
| 75 |
f'<mark style="background:{color};padding:1px 3px;border-radius:3px;" '
|
|
@@ -78,11 +100,10 @@ def highlight_text(cr_text, annotation):
|
|
| 78 |
)
|
| 79 |
pos = s["end"]
|
| 80 |
|
| 81 |
-
# Remaining text
|
| 82 |
if pos < len(cr_text):
|
| 83 |
html.append(escape_html(cr_text[pos:]))
|
| 84 |
|
| 85 |
-
return f"<pre style='white-space:pre-wrap;line-height:1.6;'>{'
|
| 86 |
|
| 87 |
|
| 88 |
def format_table(annotation):
|
|
@@ -99,8 +120,10 @@ def format_table(annotation):
|
|
| 99 |
var_label = var_name.replace("_", " ").title()
|
| 100 |
|
| 101 |
if value:
|
| 102 |
-
|
| 103 |
-
|
|
|
|
|
|
|
| 104 |
display_value = "/"
|
| 105 |
display_span = ""
|
| 106 |
else:
|
|
@@ -126,26 +149,30 @@ def format_table(annotation):
|
|
| 126 |
</table>"""
|
| 127 |
|
| 128 |
|
| 129 |
-
def display_sample(
|
| 130 |
"""Display a sample."""
|
| 131 |
-
|
| 132 |
-
if
|
| 133 |
return "Invalid index", "Invalid index"
|
| 134 |
|
| 135 |
-
|
|
|
|
| 136 |
cr = sample.get("CR", "")
|
| 137 |
annotation = sample.get("annotation", {})
|
| 138 |
|
| 139 |
-
|
|
|
|
|
|
|
|
|
|
| 140 |
|
| 141 |
|
| 142 |
# Build UI
|
| 143 |
with gr.Blocks(title="Pancreas Annotations", theme=gr.themes.Base()) as demo:
|
| 144 |
gr.Markdown("# 🔬 Pancreas Cancer Annotations Explorer")
|
| 145 |
-
gr.Markdown("
|
| 146 |
|
| 147 |
with gr.Row():
|
| 148 |
-
slider = gr.Slider(0, len(
|
| 149 |
|
| 150 |
with gr.Row():
|
| 151 |
with gr.Column():
|
|
|
|
| 7 |
|
| 8 |
# Load the dataset
|
| 9 |
print("Loading dataset...")
|
| 10 |
+
full_dataset = load_dataset("rntc/biomed-fr-pancreas-annotations", split="train")
|
| 11 |
+
print(f"Loaded {len(full_dataset)} samples")
|
| 12 |
+
|
| 13 |
+
# Filter: keep only samples with >= 10 real annotations
|
| 14 |
+
MIN_ANNOTATIONS = 10
|
| 15 |
+
|
| 16 |
+
def count_real_annotations(annotation):
|
| 17 |
+
"""Count real annotations (excluding 'not found' placeholders)."""
|
| 18 |
+
count = 0
|
| 19 |
+
for var_data in annotation.values():
|
| 20 |
+
if var_data and isinstance(var_data, dict):
|
| 21 |
+
value = var_data.get("value")
|
| 22 |
+
span = var_data.get("span", "")
|
| 23 |
+
if value:
|
| 24 |
+
if span and "pas de mention" in span.lower():
|
| 25 |
+
continue
|
| 26 |
+
if "not performed" in str(value).lower():
|
| 27 |
+
continue
|
| 28 |
+
count += 1
|
| 29 |
+
return count
|
| 30 |
+
|
| 31 |
+
# Filter dataset
|
| 32 |
+
filtered_indices = [
|
| 33 |
+
i for i, sample in enumerate(full_dataset)
|
| 34 |
+
if count_real_annotations(sample.get("annotation", {})) >= MIN_ANNOTATIONS
|
| 35 |
+
]
|
| 36 |
+
print(f"Filtered to {len(filtered_indices)} samples with >= {MIN_ANNOTATIONS} annotations")
|
| 37 |
|
| 38 |
# Colors for highlighting
|
| 39 |
COLORS = [
|
|
|
|
| 85 |
color_idx = 0
|
| 86 |
|
| 87 |
for s in filtered:
|
|
|
|
| 88 |
if s["start"] > pos:
|
| 89 |
html.append(escape_html(cr_text[pos:s["start"]]))
|
| 90 |
|
|
|
|
| 91 |
if s["var"] not in color_map:
|
| 92 |
color_map[s["var"]] = COLORS[color_idx % len(COLORS)]
|
| 93 |
color_idx += 1
|
| 94 |
|
|
|
|
| 95 |
color = color_map[s["var"]]
|
| 96 |
html.append(
|
| 97 |
f'<mark style="background:{color};padding:1px 3px;border-radius:3px;" '
|
|
|
|
| 100 |
)
|
| 101 |
pos = s["end"]
|
| 102 |
|
|
|
|
| 103 |
if pos < len(cr_text):
|
| 104 |
html.append(escape_html(cr_text[pos:]))
|
| 105 |
|
| 106 |
+
return f"<pre style='white-space:pre-wrap;line-height:1.6;'>{''.join(html)}</pre>"
|
| 107 |
|
| 108 |
|
| 109 |
def format_table(annotation):
|
|
|
|
| 120 |
var_label = var_name.replace("_", " ").title()
|
| 121 |
|
| 122 |
if value:
|
| 123 |
+
if span and "pas de mention" in span.lower():
|
| 124 |
+
display_value = "/"
|
| 125 |
+
display_span = ""
|
| 126 |
+
elif "not performed" in str(value).lower():
|
| 127 |
display_value = "/"
|
| 128 |
display_span = ""
|
| 129 |
else:
|
|
|
|
| 149 |
</table>"""
|
| 150 |
|
| 151 |
|
| 152 |
+
def display_sample(slider_idx):
|
| 153 |
"""Display a sample."""
|
| 154 |
+
slider_idx = int(slider_idx)
|
| 155 |
+
if slider_idx < 0 or slider_idx >= len(filtered_indices):
|
| 156 |
return "Invalid index", "Invalid index"
|
| 157 |
|
| 158 |
+
real_idx = filtered_indices[slider_idx]
|
| 159 |
+
sample = full_dataset[real_idx]
|
| 160 |
cr = sample.get("CR", "")
|
| 161 |
annotation = sample.get("annotation", {})
|
| 162 |
|
| 163 |
+
n_annotations = count_real_annotations(annotation)
|
| 164 |
+
header = f"<p><b>Sample #{real_idx}</b> — {n_annotations} annotations</p>"
|
| 165 |
+
|
| 166 |
+
return header + highlight_text(cr, annotation), format_table(annotation)
|
| 167 |
|
| 168 |
|
| 169 |
# Build UI
|
| 170 |
with gr.Blocks(title="Pancreas Annotations", theme=gr.themes.Base()) as demo:
|
| 171 |
gr.Markdown("# 🔬 Pancreas Cancer Annotations Explorer")
|
| 172 |
+
gr.Markdown(f"Showing {len(filtered_indices)} samples with >= {MIN_ANNOTATIONS} annotations. Hover over highlights to see values.")
|
| 173 |
|
| 174 |
with gr.Row():
|
| 175 |
+
slider = gr.Slider(0, len(filtered_indices) - 1, value=0, step=1, label="Sample")
|
| 176 |
|
| 177 |
with gr.Row():
|
| 178 |
with gr.Column():
|