add suport to cif
Browse files
app.py
CHANGED
|
@@ -13,37 +13,10 @@ default_file = "static/test.pdb"
|
|
| 13 |
TEMP_DIR = "static/tmp/"
|
| 14 |
os.makedirs(TEMP_DIR, exist_ok=True)
|
| 15 |
|
| 16 |
-
# # 保存全局信息以便交互
|
| 17 |
-
# structure_cache = gr.State()
|
| 18 |
-
# contact_cache = None
|
| 19 |
-
|
| 20 |
-
def read_file(file_path):
|
| 21 |
-
if file_path is None:
|
| 22 |
-
return "<b style='color:red'>未提供结构文件</b>"
|
| 23 |
-
|
| 24 |
-
try:
|
| 25 |
-
with open(file_path, "r") as f:
|
| 26 |
-
structure_str = f.read()
|
| 27 |
-
except Exception as e:
|
| 28 |
-
return f"<b style='color:red'>读取文件失败: {e}</b>"
|
| 29 |
-
|
| 30 |
-
# file_format = file_path.split(".")[-1]
|
| 31 |
-
summary = analyze_structure_combined(file_path)
|
| 32 |
-
entity_color_dict = build_entity_color_dict(list(summary.keys()))
|
| 33 |
-
|
| 34 |
-
# 缓存用于后续交互
|
| 35 |
-
structure_dict = {
|
| 36 |
-
"structure_str": structure_str,
|
| 37 |
-
"summary": summary,
|
| 38 |
-
"entity_color_dict": entity_color_dict
|
| 39 |
-
}
|
| 40 |
-
|
| 41 |
-
return structure_str, summary, entity_color_dict, structure_dict
|
| 42 |
-
|
| 43 |
|
| 44 |
def render_structure(structure_str, summary, entity_color_dict, add_label=True):
|
| 45 |
view = py3Dmol.view(width=233, height=233)
|
| 46 |
-
view.addModel(structure_str
|
| 47 |
set_default_styles(view, summary, entity_color_dict, add_label=add_label)
|
| 48 |
view.zoomTo()
|
| 49 |
return view
|
|
@@ -174,7 +147,7 @@ with gr.Blocks() as demo:
|
|
| 174 |
)
|
| 175 |
pdb_btn = gr.Button("获取结构", scale=1)
|
| 176 |
|
| 177 |
-
file_input = gr.File(label="或直接上传PDB文件", file_types=[".pdb"])
|
| 178 |
with gr.Column(scale=2):
|
| 179 |
with gr.Row():
|
| 180 |
entity_selector = gr.Dropdown(choices=[], interactive=True, multiselect=True, label="选择实体", scale=2)
|
|
|
|
| 13 |
TEMP_DIR = "static/tmp/"
|
| 14 |
os.makedirs(TEMP_DIR, exist_ok=True)
|
| 15 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 16 |
|
| 17 |
def render_structure(structure_str, summary, entity_color_dict, add_label=True):
|
| 18 |
view = py3Dmol.view(width=233, height=233)
|
| 19 |
+
view.addModel(structure_str) # 不指定类型似乎可以自动识别
|
| 20 |
set_default_styles(view, summary, entity_color_dict, add_label=add_label)
|
| 21 |
view.zoomTo()
|
| 22 |
return view
|
|
|
|
| 147 |
)
|
| 148 |
pdb_btn = gr.Button("获取结构", scale=1)
|
| 149 |
|
| 150 |
+
file_input = gr.File(label="或直接上传PDB文件", file_types=[".pdb", ".cif", ".ent"])
|
| 151 |
with gr.Column(scale=2):
|
| 152 |
with gr.Row():
|
| 153 |
entity_selector = gr.Dropdown(choices=[], interactive=True, multiselect=True, label="选择实体", scale=2)
|
utils.py
CHANGED
|
@@ -4,13 +4,13 @@ import hashlib
|
|
| 4 |
import numpy as np
|
| 5 |
|
| 6 |
from tempfile import NamedTemporaryFile
|
| 7 |
-
|
| 8 |
-
from Bio.PDB import MMCIFParser, PDBParser,
|
| 9 |
from collections import defaultdict
|
| 10 |
import colorsys
|
| 11 |
import hashlib
|
| 12 |
from Bio.PDB.NeighborSearch import NeighborSearch
|
| 13 |
-
|
| 14 |
|
| 15 |
NUCLEIC_ACIDS = {
|
| 16 |
"A", "G", "C", "U", "T",
|
|
@@ -112,7 +112,17 @@ def classify_residue(residue):
|
|
| 112 |
|
| 113 |
def analyze_structure_combined(file_path):
|
| 114 |
if file_path.endswith(".cif"):
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 115 |
parser = MMCIFParser(QUIET=True)
|
|
|
|
| 116 |
elif file_path.endswith(".pdb") or file_path.endswith(".ent"):
|
| 117 |
parser = PDBParser(QUIET=True)
|
| 118 |
else:
|
|
@@ -149,6 +159,30 @@ def analyze_structure_combined(file_path):
|
|
| 149 |
))
|
| 150 |
return sorted_summary
|
| 151 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 152 |
def extract_contact_residues(summary, selected_keys, cutoff=3.5):
|
| 153 |
entity_atoms = {key: [] for key in selected_keys}
|
| 154 |
atom_to_residue_info = {}
|
|
@@ -247,7 +281,7 @@ def extract_polar_contacts(summary, contact_summary, cutoff=3.5, angle_cutoff=12
|
|
| 247 |
|
| 248 |
def set_default_styles(viewer, summary, entity_color_dict,
|
| 249 |
add_label=True):
|
| 250 |
-
|
| 251 |
for entity, color in entity_color_dict.items():
|
| 252 |
label_style = {
|
| 253 |
'fontOpacity':1,
|
|
@@ -260,18 +294,19 @@ def set_default_styles(viewer, summary, entity_color_dict,
|
|
| 260 |
if "(protein)" in entity.lower():
|
| 261 |
# 提取链ID
|
| 262 |
chain_id = entity.split()[1]
|
| 263 |
-
|
| 264 |
-
|
| 265 |
-
|
|
|
|
| 266 |
if add_label:
|
| 267 |
viewer.addLabel(entity, label_style, {'chain':chain_id})
|
| 268 |
elif '(dna)' in entity.lower() or '(rna)' in entity.lower():
|
| 269 |
# 提取链ID
|
| 270 |
chain_id = entity.split()[1]
|
| 271 |
-
if chain_id == 'T':
|
| 272 |
-
print(summary[entity])
|
| 273 |
viewer.setStyle({'chain': chain_id},
|
| 274 |
-
{'cartoon': {'color':
|
|
|
|
|
|
|
| 275 |
if add_label:
|
| 276 |
viewer.addLabel(entity, label_style, {'chain':chain_id})
|
| 277 |
|
|
|
|
| 4 |
import numpy as np
|
| 5 |
|
| 6 |
from tempfile import NamedTemporaryFile
|
| 7 |
+
from io import StringIO
|
| 8 |
+
from Bio.PDB import MMCIFParser, PDBParser, PDBIO
|
| 9 |
from collections import defaultdict
|
| 10 |
import colorsys
|
| 11 |
import hashlib
|
| 12 |
from Bio.PDB.NeighborSearch import NeighborSearch
|
| 13 |
+
from Bio.PDB.DSSP import DSSP
|
| 14 |
|
| 15 |
NUCLEIC_ACIDS = {
|
| 16 |
"A", "G", "C", "U", "T",
|
|
|
|
| 112 |
|
| 113 |
def analyze_structure_combined(file_path):
|
| 114 |
if file_path.endswith(".cif"):
|
| 115 |
+
with open(file_path, 'r') as f:
|
| 116 |
+
content = f.read()
|
| 117 |
+
# 如果缺少 data_ 开头,就加上一个默认块名
|
| 118 |
+
if not content.lstrip().startswith("data_"):
|
| 119 |
+
content = "data_auto\n" + content
|
| 120 |
+
# 3. 写入临时 mmCIF 文件
|
| 121 |
+
with NamedTemporaryFile(suffix=".cif", delete=False, mode='w') as tmp:
|
| 122 |
+
tmp.write(content)
|
| 123 |
+
file_path = tmp.name
|
| 124 |
parser = MMCIFParser(QUIET=True)
|
| 125 |
+
|
| 126 |
elif file_path.endswith(".pdb") or file_path.endswith(".ent"):
|
| 127 |
parser = PDBParser(QUIET=True)
|
| 128 |
else:
|
|
|
|
| 159 |
))
|
| 160 |
return sorted_summary
|
| 161 |
|
| 162 |
+
def read_file(file_path):
|
| 163 |
+
if file_path is None:
|
| 164 |
+
return "<b style='color:red'>未提供结构文件</b>"
|
| 165 |
+
|
| 166 |
+
try:
|
| 167 |
+
with open(file_path, "r") as f:
|
| 168 |
+
structure_str = f.read()
|
| 169 |
+
except Exception as e:
|
| 170 |
+
return f"<b style='color:red'>读取文件失败: {e}</b>"
|
| 171 |
+
|
| 172 |
+
# file_format = file_path.split(".")[-1]
|
| 173 |
+
summary = analyze_structure_combined(file_path)
|
| 174 |
+
entity_color_dict = build_entity_color_dict(list(summary.keys()))
|
| 175 |
+
|
| 176 |
+
# 缓存用于后续交互
|
| 177 |
+
structure_dict = {
|
| 178 |
+
"structure_str": structure_str,
|
| 179 |
+
"summary": summary,
|
| 180 |
+
"entity_color_dict": entity_color_dict
|
| 181 |
+
}
|
| 182 |
+
|
| 183 |
+
return structure_str, summary, entity_color_dict, structure_dict
|
| 184 |
+
|
| 185 |
+
|
| 186 |
def extract_contact_residues(summary, selected_keys, cutoff=3.5):
|
| 187 |
entity_atoms = {key: [] for key in selected_keys}
|
| 188 |
atom_to_residue_info = {}
|
|
|
|
| 281 |
|
| 282 |
def set_default_styles(viewer, summary, entity_color_dict,
|
| 283 |
add_label=True):
|
| 284 |
+
viewer.setStyle({'hetflag': True}, {"stick": {}})
|
| 285 |
for entity, color in entity_color_dict.items():
|
| 286 |
label_style = {
|
| 287 |
'fontOpacity':1,
|
|
|
|
| 294 |
if "(protein)" in entity.lower():
|
| 295 |
# 提取链ID
|
| 296 |
chain_id = entity.split()[1]
|
| 297 |
+
viewer.setStyle({'chain': chain_id},
|
| 298 |
+
{'cartoon': {'arrows': True,
|
| 299 |
+
'color': color,
|
| 300 |
+
'opacity': 0.9}})
|
| 301 |
if add_label:
|
| 302 |
viewer.addLabel(entity, label_style, {'chain':chain_id})
|
| 303 |
elif '(dna)' in entity.lower() or '(rna)' in entity.lower():
|
| 304 |
# 提取链ID
|
| 305 |
chain_id = entity.split()[1]
|
|
|
|
|
|
|
| 306 |
viewer.setStyle({'chain': chain_id},
|
| 307 |
+
{'cartoon': {'color': color,
|
| 308 |
+
'nucleicAcid': True,
|
| 309 |
+
'opacity': 0.8}}) # 碱基和磷酸用stick)
|
| 310 |
if add_label:
|
| 311 |
viewer.addLabel(entity, label_style, {'chain':chain_id})
|
| 312 |
|