Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
|
@@ -5,7 +5,7 @@ import numpy as np
|
|
| 5 |
from itertools import product
|
| 6 |
import torch.nn as nn
|
| 7 |
import matplotlib.pyplot as plt
|
| 8 |
-
import matplotlib.colors as mcolors
|
| 9 |
import io
|
| 10 |
from io import BytesIO # Import io then BytesIO
|
| 11 |
from PIL import Image, ImageDraw, ImageFont
|
|
@@ -16,6 +16,9 @@ import pandas as pd
|
|
| 16 |
import tempfile
|
| 17 |
import os
|
| 18 |
from typing import List, Dict, Tuple, Optional, Any
|
|
|
|
|
|
|
|
|
|
| 19 |
|
| 20 |
###############################################################################
|
| 21 |
# 1. MODEL DEFINITION
|
|
@@ -567,18 +570,6 @@ def analyze_sequence_comparison(file1, file2, fasta1="", fasta2=""):
|
|
| 567 |
# 11. GENE FEATURE ANALYSIS
|
| 568 |
###############################################################################
|
| 569 |
|
| 570 |
-
import io
|
| 571 |
-
from io import BytesIO
|
| 572 |
-
from PIL import Image, ImageDraw, ImageFont
|
| 573 |
-
import numpy as np
|
| 574 |
-
import pandas as pd
|
| 575 |
-
import tempfile
|
| 576 |
-
import os
|
| 577 |
-
from typing import List, Dict, Tuple, Optional, Any
|
| 578 |
-
import matplotlib.pyplot as plt
|
| 579 |
-
from matplotlib.colors import LinearSegmentedColormap
|
| 580 |
-
import seaborn as sns
|
| 581 |
-
|
| 582 |
def parse_gene_features(text: str) -> List[Dict[str, Any]]:
|
| 583 |
"""Parse gene features from text file in FASTA-like format"""
|
| 584 |
genes = []
|
|
@@ -653,6 +644,19 @@ def compute_gene_statistics(gene_shap: np.ndarray) -> Dict[str, float]:
|
|
| 653 |
|
| 654 |
def create_simple_genome_diagram(gene_results: List[Dict[str, Any]], genome_length: int) -> Image.Image:
|
| 655 |
"""Create a simple genome diagram using PIL"""
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 656 |
# Image dimensions
|
| 657 |
width = 1500
|
| 658 |
height = 600
|
|
@@ -684,7 +688,7 @@ def create_simple_genome_diagram(gene_results: List[Dict[str, Any]], genome_leng
|
|
| 684 |
|
| 685 |
# Draw scale markers
|
| 686 |
for i in range(0, genome_length + 1, genome_length // 10):
|
| 687 |
-
x = margin + i * scale
|
| 688 |
draw.line([(x, line_y - 5), (x, line_y + 5)], fill='black', width=1)
|
| 689 |
draw.text((x - 20, line_y + 10), f"{i:,}", fill='black', font=font)
|
| 690 |
|
|
@@ -693,9 +697,9 @@ def create_simple_genome_diagram(gene_results: List[Dict[str, Any]], genome_leng
|
|
| 693 |
|
| 694 |
# Draw genes
|
| 695 |
for gene in sorted_genes:
|
| 696 |
-
# Calculate position
|
| 697 |
-
start_x = margin + gene['start'] * scale
|
| 698 |
-
end_x = margin + gene['end'] * scale
|
| 699 |
|
| 700 |
# Calculate color based on SHAP value
|
| 701 |
if gene['avg_shap'] > 0:
|
|
@@ -705,10 +709,10 @@ def create_simple_genome_diagram(gene_results: List[Dict[str, Any]], genome_leng
|
|
| 705 |
intensity = min(255, int(abs(gene['avg_shap'] * 500)))
|
| 706 |
color = (255 - intensity, 255 - intensity, 255) # Blue
|
| 707 |
|
| 708 |
-
# Draw gene box
|
| 709 |
draw.rectangle([
|
| 710 |
-
(start_x, line_y - track_height // 2),
|
| 711 |
-
(end_x, line_y + track_height // 2)
|
| 712 |
], fill=color, outline='black')
|
| 713 |
|
| 714 |
# Draw gene name
|
|
@@ -726,8 +730,8 @@ def create_simple_genome_diagram(gene_results: List[Dict[str, Any]], genome_leng
|
|
| 726 |
gene_width = end_x - start_x
|
| 727 |
if gene_width > label_width:
|
| 728 |
# Horizontal label
|
| 729 |
-
text_x = start_x + (gene_width - label_width) // 2
|
| 730 |
-
draw.text((text_x, text_y), label, fill='black', font=font)
|
| 731 |
elif gene_width > 20:
|
| 732 |
# Create rotated text image
|
| 733 |
txt_img = Image.new('RGBA', (label_width, 20), (255, 255, 255, 0))
|
|
|
|
| 5 |
from itertools import product
|
| 6 |
import torch.nn as nn
|
| 7 |
import matplotlib.pyplot as plt
|
| 8 |
+
import matplotlib.colors as mcolors LinearSegmentedColormap
|
| 9 |
import io
|
| 10 |
from io import BytesIO # Import io then BytesIO
|
| 11 |
from PIL import Image, ImageDraw, ImageFont
|
|
|
|
| 16 |
import tempfile
|
| 17 |
import os
|
| 18 |
from typing import List, Dict, Tuple, Optional, Any
|
| 19 |
+
import io
|
| 20 |
+
from io import BytesIO
|
| 21 |
+
import seaborn as sns
|
| 22 |
|
| 23 |
###############################################################################
|
| 24 |
# 1. MODEL DEFINITION
|
|
|
|
| 570 |
# 11. GENE FEATURE ANALYSIS
|
| 571 |
###############################################################################
|
| 572 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 573 |
def parse_gene_features(text: str) -> List[Dict[str, Any]]:
|
| 574 |
"""Parse gene features from text file in FASTA-like format"""
|
| 575 |
genes = []
|
|
|
|
| 644 |
|
| 645 |
def create_simple_genome_diagram(gene_results: List[Dict[str, Any]], genome_length: int) -> Image.Image:
|
| 646 |
"""Create a simple genome diagram using PIL"""
|
| 647 |
+
# Validate inputs
|
| 648 |
+
if not gene_results or genome_length <= 0:
|
| 649 |
+
img = Image.new('RGB', (800, 100), color='white')
|
| 650 |
+
draw = ImageDraw.Draw(img)
|
| 651 |
+
draw.text((10, 40), "Error: Invalid input data", fill='black')
|
| 652 |
+
return img
|
| 653 |
+
|
| 654 |
+
# Ensure all gene coordinates are valid integers
|
| 655 |
+
for gene in gene_results:
|
| 656 |
+
gene['start'] = max(0, int(gene['start']))
|
| 657 |
+
gene['end'] = min(genome_length, int(gene['end']))
|
| 658 |
+
if gene['start'] >= gene['end']:
|
| 659 |
+
print(f"Warning: Invalid coordinates for gene {gene['gene_name']}: {gene['start']}-{gene['end']}")
|
| 660 |
# Image dimensions
|
| 661 |
width = 1500
|
| 662 |
height = 600
|
|
|
|
| 688 |
|
| 689 |
# Draw scale markers
|
| 690 |
for i in range(0, genome_length + 1, genome_length // 10):
|
| 691 |
+
x = int(margin + i * scale)
|
| 692 |
draw.line([(x, line_y - 5), (x, line_y + 5)], fill='black', width=1)
|
| 693 |
draw.text((x - 20, line_y + 10), f"{i:,}", fill='black', font=font)
|
| 694 |
|
|
|
|
| 697 |
|
| 698 |
# Draw genes
|
| 699 |
for gene in sorted_genes:
|
| 700 |
+
# Calculate position and ensure integers
|
| 701 |
+
start_x = int(margin + gene['start'] * scale)
|
| 702 |
+
end_x = int(margin + gene['end'] * scale)
|
| 703 |
|
| 704 |
# Calculate color based on SHAP value
|
| 705 |
if gene['avg_shap'] > 0:
|
|
|
|
| 709 |
intensity = min(255, int(abs(gene['avg_shap'] * 500)))
|
| 710 |
color = (255 - intensity, 255 - intensity, 255) # Blue
|
| 711 |
|
| 712 |
+
# Draw gene box with integer coordinates
|
| 713 |
draw.rectangle([
|
| 714 |
+
(start_x, int(line_y - track_height // 2)),
|
| 715 |
+
(end_x, int(line_y + track_height // 2))
|
| 716 |
], fill=color, outline='black')
|
| 717 |
|
| 718 |
# Draw gene name
|
|
|
|
| 730 |
gene_width = end_x - start_x
|
| 731 |
if gene_width > label_width:
|
| 732 |
# Horizontal label
|
| 733 |
+
text_x = int(start_x + (gene_width - label_width) // 2)
|
| 734 |
+
draw.text((text_x, int(text_y)), label, fill='black', font=font)
|
| 735 |
elif gene_width > 20:
|
| 736 |
# Create rotated text image
|
| 737 |
txt_img = Image.new('RGBA', (label_width, 20), (255, 255, 255, 0))
|