Spaces:
Sleeping
Sleeping
Upload 2 files
Browse files- app.py +175 -14
- requirements.txt +2 -2
app.py
CHANGED
|
@@ -1,13 +1,38 @@
|
|
| 1 |
import gradio as gr # type: ignore
|
| 2 |
import os
|
|
|
|
| 3 |
|
| 4 |
# Import your custom modules from the /scripts folder
|
| 5 |
from scripts.download import download_and_clean_pdb
|
| 6 |
from scripts.generator import run_broteinshake_generator
|
| 7 |
from scripts.refine import polish_design
|
|
|
|
| 8 |
|
| 9 |
# --- HELPER FUNCTIONS ---
|
| 10 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 11 |
def extract_best_sequence(fasta_file: str) -> str:
|
| 12 |
"""Extract the best sequence (lowest score) from FASTA file."""
|
| 13 |
best_score = float('inf')
|
|
@@ -54,17 +79,32 @@ def run_part1(pdb_id, fixed_chains, variable_chains):
|
|
| 54 |
# This creates the .fa files you need for the ESM Atlas
|
| 55 |
run_broteinshake_generator(pdb_path, fixed_chains, variable_chains)
|
| 56 |
|
| 57 |
-
#
|
| 58 |
fa_file = os.path.join("generated", pdb_id.lower(), "seqs", f"{pdb_id.lower()}_clones.fa")
|
|
|
|
| 59 |
best_sequence = extract_best_sequence(fa_file)
|
| 60 |
|
|
|
|
|
|
|
|
|
|
| 61 |
# Parse score from header for status message
|
| 62 |
score_part = [p for p in best_sequence.split('\n')[0].split(',') if 'score' in p][0]
|
| 63 |
best_score = float(score_part.split('=')[1])
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 64 |
|
| 65 |
-
return
|
| 66 |
except Exception as e:
|
| 67 |
-
return "", f"Error in Part 1: {str(e)}"
|
| 68 |
|
| 69 |
def run_part2(pdb_id, uploaded_esm_file):
|
| 70 |
"""Aligns the ESM-folded PDB to the target structure."""
|
|
@@ -72,8 +112,27 @@ def run_part2(pdb_id, uploaded_esm_file):
|
|
| 72 |
if uploaded_esm_file is None:
|
| 73 |
return None, "β οΈ Please upload the PDB from ESM Atlas first."
|
| 74 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 75 |
# Call the new lightweight Biopython-only script
|
| 76 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 77 |
|
| 78 |
# Update the report to focus on Backbone Alignment
|
| 79 |
report = (
|
|
@@ -82,12 +141,61 @@ def run_part2(pdb_id, uploaded_esm_file):
|
|
| 82 |
f"𧬠Status: High-Precision Backbone Match"
|
| 83 |
)
|
| 84 |
return final_pdb_path, report
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 85 |
except Exception as e:
|
| 86 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 87 |
|
| 88 |
# --- GRADIO INTERFACE ---
|
| 89 |
|
| 90 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 91 |
gr.Markdown("# broteinshake")
|
| 92 |
gr.Markdown("### AI-Driven Structural Mimicry & BBB-Shuttle Design")
|
| 93 |
|
|
@@ -100,24 +208,77 @@ with gr.Blocks(theme=gr.themes.Soft()) as demo:
|
|
| 100 |
f_chains = gr.Textbox(label="Fixed Chains (Lock)", value="A")
|
| 101 |
v_chains = gr.Textbox(label="Variable Chains (Key)", value="B")
|
| 102 |
|
| 103 |
-
gen_btn = gr.Button("Generate Optimized Sequences", variant="primary")
|
| 104 |
-
|
| 105 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 106 |
|
| 107 |
-
gen_btn.click(run_part1, inputs=[pdb_input, f_chains, v_chains],
|
|
|
|
| 108 |
|
| 109 |
# TAB 2: STRUCTURAL VALIDATION
|
| 110 |
with gr.Tab("2. Structural Validation"):
|
| 111 |
-
gr.Markdown("###
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 112 |
with gr.Row():
|
| 113 |
esm_upload = gr.File(label="Upload ESM-Folded PDB", file_types=[".pdb"])
|
| 114 |
refined_download = gr.File(label="Download Aligned Lead (.pdb)")
|
| 115 |
|
| 116 |
-
# Change the button text to match our new approach
|
| 117 |
validate_btn = gr.Button("β¨ Run Structural Alignment", variant="primary")
|
| 118 |
status2 = gr.Textbox(label="Validation Report", interactive=False, lines=5)
|
| 119 |
-
|
| 120 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 121 |
|
| 122 |
# Launch the app
|
| 123 |
if __name__ == "__main__":
|
|
|
|
| 1 |
import gradio as gr # type: ignore
|
| 2 |
import os
|
| 3 |
+
from gradio_molecule3d import Molecule3D
|
| 4 |
|
| 5 |
# Import your custom modules from the /scripts folder
|
| 6 |
from scripts.download import download_and_clean_pdb
|
| 7 |
from scripts.generator import run_broteinshake_generator
|
| 8 |
from scripts.refine import polish_design
|
| 9 |
+
from scripts.visualize import create_design_plot
|
| 10 |
|
| 11 |
# --- HELPER FUNCTIONS ---
|
| 12 |
|
| 13 |
+
def get_all_sequences(fasta_file: str) -> str:
|
| 14 |
+
"""Get all designed sequences from FASTA file."""
|
| 15 |
+
sequences = []
|
| 16 |
+
with open(fasta_file, 'r') as f:
|
| 17 |
+
lines = [line.strip() for line in f.readlines() if line.strip()]
|
| 18 |
+
|
| 19 |
+
for i in range(0, len(lines), 2):
|
| 20 |
+
if i + 1 >= len(lines):
|
| 21 |
+
break
|
| 22 |
+
header = lines[i]
|
| 23 |
+
sequence = lines[i+1]
|
| 24 |
+
|
| 25 |
+
# Skip the original native sequence (first entry)
|
| 26 |
+
if "sample" not in header:
|
| 27 |
+
continue
|
| 28 |
+
|
| 29 |
+
sequences.append(f"{header}\n{sequence}")
|
| 30 |
+
|
| 31 |
+
if sequences:
|
| 32 |
+
return "\n\n".join(sequences)
|
| 33 |
+
else:
|
| 34 |
+
raise ValueError(f"No valid designs found in {fasta_file}")
|
| 35 |
+
|
| 36 |
def extract_best_sequence(fasta_file: str) -> str:
|
| 37 |
"""Extract the best sequence (lowest score) from FASTA file."""
|
| 38 |
best_score = float('inf')
|
|
|
|
| 79 |
# This creates the .fa files you need for the ESM Atlas
|
| 80 |
run_broteinshake_generator(pdb_path, fixed_chains, variable_chains)
|
| 81 |
|
| 82 |
+
# Get all sequences and the best one
|
| 83 |
fa_file = os.path.join("generated", pdb_id.lower(), "seqs", f"{pdb_id.lower()}_clones.fa")
|
| 84 |
+
all_sequences = get_all_sequences(fa_file)
|
| 85 |
best_sequence = extract_best_sequence(fa_file)
|
| 86 |
|
| 87 |
+
# Generate the dashboard plot
|
| 88 |
+
evolution_plot = create_design_plot(fa_file)
|
| 89 |
+
|
| 90 |
# Parse score from header for status message
|
| 91 |
score_part = [p for p in best_sequence.split('\n')[0].split(',') if 'score' in p][0]
|
| 92 |
best_score = float(score_part.split('=')[1])
|
| 93 |
+
|
| 94 |
+
# Count number of designs
|
| 95 |
+
num_designs = len([s for s in all_sequences.split('\n\n') if s.strip()])
|
| 96 |
+
|
| 97 |
+
# Format status with best sequence
|
| 98 |
+
status_message = (
|
| 99 |
+
f"β
Design Complete! {num_designs} designs generated.\n\n"
|
| 100 |
+
f"β Lead Candidate (Best Score: {best_score:.4f}):\n"
|
| 101 |
+
f"{best_sequence}\n\n"
|
| 102 |
+
f"π Copy the lead sequence above and fold it at [esmatlas.com](https://esmatlas.com/resources?action=fold)."
|
| 103 |
+
)
|
| 104 |
|
| 105 |
+
return all_sequences, evolution_plot, status_message
|
| 106 |
except Exception as e:
|
| 107 |
+
return "", None, f"β Error in Part 1: {str(e)}"
|
| 108 |
|
| 109 |
def run_part2(pdb_id, uploaded_esm_file):
|
| 110 |
"""Aligns the ESM-folded PDB to the target structure."""
|
|
|
|
| 112 |
if uploaded_esm_file is None:
|
| 113 |
return None, "β οΈ Please upload the PDB from ESM Atlas first."
|
| 114 |
|
| 115 |
+
# Validate inputs
|
| 116 |
+
if not pdb_id or not pdb_id.strip():
|
| 117 |
+
return None, "β Error: PDB ID is required."
|
| 118 |
+
|
| 119 |
+
# Get file path
|
| 120 |
+
file_path = uploaded_esm_file.name if hasattr(uploaded_esm_file, 'name') else uploaded_esm_file
|
| 121 |
+
if not file_path or not os.path.exists(file_path):
|
| 122 |
+
return None, f"β Error: Uploaded file not found: {file_path}"
|
| 123 |
+
|
| 124 |
# Call the new lightweight Biopython-only script
|
| 125 |
+
print(f"π Starting alignment for PDB ID: {pdb_id}, File: {file_path}")
|
| 126 |
+
final_pdb_path, rmsd_val = polish_design(pdb_id, file_path)
|
| 127 |
+
|
| 128 |
+
# Validate output
|
| 129 |
+
if not final_pdb_path or not os.path.exists(final_pdb_path):
|
| 130 |
+
return None, f"β Error: Alignment failed - output file not created: {final_pdb_path}"
|
| 131 |
+
|
| 132 |
+
if rmsd_val is None:
|
| 133 |
+
return None, "β Error: Alignment failed - RMSD calculation returned None"
|
| 134 |
+
|
| 135 |
+
print(f"β
Alignment successful: RMSD = {rmsd_val:.3f} Γ
, Output: {final_pdb_path}")
|
| 136 |
|
| 137 |
# Update the report to focus on Backbone Alignment
|
| 138 |
report = (
|
|
|
|
| 141 |
f"𧬠Status: High-Precision Backbone Match"
|
| 142 |
)
|
| 143 |
return final_pdb_path, report
|
| 144 |
+
except FileNotFoundError as e:
|
| 145 |
+
error_msg = f"β File Error: {str(e)}"
|
| 146 |
+
print(error_msg)
|
| 147 |
+
return None, error_msg
|
| 148 |
+
except ValueError as e:
|
| 149 |
+
error_msg = f"β Validation Error: {str(e)}"
|
| 150 |
+
print(error_msg)
|
| 151 |
+
return None, error_msg
|
| 152 |
except Exception as e:
|
| 153 |
+
error_msg = f"β Unexpected Error: {str(e)}\n\nTraceback:\n{type(e).__name__}"
|
| 154 |
+
print(error_msg)
|
| 155 |
+
import traceback
|
| 156 |
+
traceback.print_exc()
|
| 157 |
+
return None, error_msg
|
| 158 |
|
| 159 |
# --- GRADIO INTERFACE ---
|
| 160 |
|
| 161 |
+
# 1. Define the Biohub Theme
|
| 162 |
+
biohub_theme = gr.themes.Soft(
|
| 163 |
+
primary_hue="emerald",
|
| 164 |
+
secondary_hue="slate",
|
| 165 |
+
neutral_hue="slate",
|
| 166 |
+
font=[gr.themes.GoogleFont("Inter"), "ui-sans-serif", "system-ui"],
|
| 167 |
+
).set(
|
| 168 |
+
# Custom CSS variable overrides for that "Glass" look
|
| 169 |
+
block_title_text_weight="600",
|
| 170 |
+
block_label_text_size="*text_sm",
|
| 171 |
+
input_background_fill="*neutral_50",
|
| 172 |
+
button_primary_background_fill="linear-gradient(90deg, #10b981, #34d399)",
|
| 173 |
+
button_primary_text_color="white",
|
| 174 |
+
border_color_primary="*neutral_200",
|
| 175 |
+
)
|
| 176 |
+
|
| 177 |
+
# 2. Inject Custom CSS for the "Spice"
|
| 178 |
+
biohub_css = """
|
| 179 |
+
.gradio-container {
|
| 180 |
+
background: radial-gradient(circle at top right, #f8fafc, #f1f5f9); /* Subtle depth */
|
| 181 |
+
}
|
| 182 |
+
#biohub-header {
|
| 183 |
+
text-align: center;
|
| 184 |
+
padding: 20px;
|
| 185 |
+
background: #064e3b;
|
| 186 |
+
color: white;
|
| 187 |
+
border-radius: 15px;
|
| 188 |
+
margin-bottom: 20px;
|
| 189 |
+
box-shadow: 0 4px 6px -1px rgba(0, 0, 0, 0.1);
|
| 190 |
+
}
|
| 191 |
+
.gr-box {
|
| 192 |
+
border-radius: 12px !important;
|
| 193 |
+
border: 1px solid #e2e8f0 !important;
|
| 194 |
+
}
|
| 195 |
+
footer {display: none !important;} /* Cleans up the portfolio look */
|
| 196 |
+
"""
|
| 197 |
+
|
| 198 |
+
with gr.Blocks(theme=biohub_theme, css=biohub_css) as demo:
|
| 199 |
gr.Markdown("# broteinshake")
|
| 200 |
gr.Markdown("### AI-Driven Structural Mimicry & BBB-Shuttle Design")
|
| 201 |
|
|
|
|
| 208 |
f_chains = gr.Textbox(label="Fixed Chains (Lock)", value="A")
|
| 209 |
v_chains = gr.Textbox(label="Variable Chains (Key)", value="B")
|
| 210 |
|
| 211 |
+
gen_btn = gr.Button("π Generate Optimized Sequences", variant="primary")
|
| 212 |
+
|
| 213 |
+
# Stack components vertically
|
| 214 |
+
fa_output = gr.Code(
|
| 215 |
+
label="Designed Sequences",
|
| 216 |
+
language="markdown",
|
| 217 |
+
lines=10,
|
| 218 |
+
max_lines=20 # Prevents infinite growth when showing all 20 candidates
|
| 219 |
+
)
|
| 220 |
+
plot_output = gr.Plot(
|
| 221 |
+
label="Design Evolution Dashboard"
|
| 222 |
+
)
|
| 223 |
+
|
| 224 |
+
gr.Markdown("### System Status")
|
| 225 |
+
status1 = gr.Markdown()
|
| 226 |
|
| 227 |
+
gen_btn.click(run_part1, inputs=[pdb_input, f_chains, v_chains],
|
| 228 |
+
outputs=[fa_output, plot_output, status1])
|
| 229 |
|
| 230 |
# TAB 2: STRUCTURAL VALIDATION
|
| 231 |
with gr.Tab("2. Structural Validation"):
|
| 232 |
+
gr.Markdown("### 𧬠Final Structure Preview")
|
| 233 |
+
|
| 234 |
+
# Updated REPS for local dev visibility
|
| 235 |
+
REPS = [
|
| 236 |
+
{
|
| 237 |
+
"model": 0,
|
| 238 |
+
"style": "cartoon",
|
| 239 |
+
"color": "spectrum",
|
| 240 |
+
"opacity": 1.0
|
| 241 |
+
# Removing "chain": "A" is the key fix here!
|
| 242 |
+
}
|
| 243 |
+
]
|
| 244 |
+
|
| 245 |
+
# This component ACTUALLY supports .pdb natively
|
| 246 |
+
protein_view = Molecule3D(label="3D Structure Viewer (Refined Shuttle)", reps=REPS)
|
| 247 |
+
|
| 248 |
with gr.Row():
|
| 249 |
esm_upload = gr.File(label="Upload ESM-Folded PDB", file_types=[".pdb"])
|
| 250 |
refined_download = gr.File(label="Download Aligned Lead (.pdb)")
|
| 251 |
|
|
|
|
| 252 |
validate_btn = gr.Button("β¨ Run Structural Alignment", variant="primary")
|
| 253 |
status2 = gr.Textbox(label="Validation Report", interactive=False, lines=5)
|
| 254 |
+
|
| 255 |
+
# Update viewer - REPS are set on component creation, just return the file path
|
| 256 |
+
def run_validation_with_view(pdb_id, file):
|
| 257 |
+
try:
|
| 258 |
+
# This is your existing alignment/refine logic
|
| 259 |
+
final_pdb, report = run_part2(pdb_id, file)
|
| 260 |
+
|
| 261 |
+
if final_pdb is not None and os.path.exists(final_pdb):
|
| 262 |
+
# Verify we're using the refined shuttle (Refined_Shuttle.pdb)
|
| 263 |
+
if "Refined_Shuttle.pdb" in final_pdb or os.path.basename(final_pdb) == "Refined_Shuttle.pdb":
|
| 264 |
+
print(f"π― Visualizing refined shuttle: {final_pdb}")
|
| 265 |
+
else:
|
| 266 |
+
print(f"β οΈ Warning: Expected Refined_Shuttle.pdb but got: {final_pdb}")
|
| 267 |
+
|
| 268 |
+
# Molecule3D expects just the file path string (REPS are already set on component)
|
| 269 |
+
return final_pdb, report, final_pdb
|
| 270 |
+
except Exception as e:
|
| 271 |
+
error_msg = f"β Error generating 3D view: {str(e)}"
|
| 272 |
+
print(error_msg)
|
| 273 |
+
import traceback
|
| 274 |
+
traceback.print_exc()
|
| 275 |
+
return None, error_msg, None
|
| 276 |
+
|
| 277 |
+
validate_btn.click(
|
| 278 |
+
run_validation_with_view,
|
| 279 |
+
inputs=[pdb_input, esm_upload],
|
| 280 |
+
outputs=[refined_download, status2, protein_view]
|
| 281 |
+
)
|
| 282 |
|
| 283 |
# Launch the app
|
| 284 |
if __name__ == "__main__":
|
requirements.txt
CHANGED
|
@@ -1,4 +1,5 @@
|
|
| 1 |
-
gradio
|
|
|
|
| 2 |
accelerate==1.12.0
|
| 3 |
backports.zstd
|
| 4 |
biopython==1.86
|
|
@@ -19,7 +20,6 @@ MarkupSafe
|
|
| 19 |
mpmath
|
| 20 |
networkx
|
| 21 |
numpy==2.2.6
|
| 22 |
-
openmm
|
| 23 |
packaging==25.0
|
| 24 |
pillow
|
| 25 |
psutil==7.2.1
|
|
|
|
| 1 |
+
gradio==5.50.0
|
| 2 |
+
gradio_molecule3d
|
| 3 |
accelerate==1.12.0
|
| 4 |
backports.zstd
|
| 5 |
biopython==1.86
|
|
|
|
| 20 |
mpmath
|
| 21 |
networkx
|
| 22 |
numpy==2.2.6
|
|
|
|
| 23 |
packaging==25.0
|
| 24 |
pillow
|
| 25 |
psutil==7.2.1
|