Spaces:
Sleeping
Sleeping
fix train.py
Browse files
train.py
ADDED
@@ -0,0 +1,611 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from __future__ import annotations
|
2 |
+
import time
|
3 |
+
import json
|
4 |
+
import gradio as gr
|
5 |
+
from gradio_molecule3d import Molecule3D
|
6 |
+
import torch
|
7 |
+
from pinder.core import get_pinder_location
|
8 |
+
get_pinder_location()
|
9 |
+
from pytorch_lightning import LightningModule
|
10 |
+
|
11 |
+
import torch
|
12 |
+
import lightning.pytorch as pl
|
13 |
+
import torch.nn.functional as F
|
14 |
+
|
15 |
+
import torch.nn as nn
|
16 |
+
import torchmetrics
|
17 |
+
import torch.nn as nn
|
18 |
+
import torch.nn.functional as F
|
19 |
+
from torch_geometric.nn import MessagePassing
|
20 |
+
from torch_geometric.nn import global_mean_pool
|
21 |
+
from torch.nn import Sequential, Linear, BatchNorm1d, ReLU
|
22 |
+
from torch_scatter import scatter
|
23 |
+
from torch.nn import Module
|
24 |
+
|
25 |
+
|
26 |
+
import pinder.core as pinder
|
27 |
+
pinder.__version__
|
28 |
+
from torch_geometric.loader import DataLoader
|
29 |
+
from pinder.core.loader.dataset import get_geo_loader
|
30 |
+
from pinder.core import download_dataset
|
31 |
+
from pinder.core import get_index
|
32 |
+
from pinder.core import get_metadata
|
33 |
+
from pathlib import Path
|
34 |
+
import pandas as pd
|
35 |
+
from pinder.core import PinderSystem
|
36 |
+
import torch
|
37 |
+
from pinder.core.loader.dataset import PPIDataset
|
38 |
+
from pinder.core.loader.geodata import NodeRepresentation
|
39 |
+
import pickle
|
40 |
+
from pinder.core import get_index, PinderSystem
|
41 |
+
from torch_geometric.data import HeteroData
|
42 |
+
import os
|
43 |
+
|
44 |
+
from enum import Enum
|
45 |
+
|
46 |
+
import numpy as np
|
47 |
+
import torch
|
48 |
+
import lightning.pytorch as pl
|
49 |
+
from numpy.typing import NDArray
|
50 |
+
from torch_geometric.data import HeteroData
|
51 |
+
|
52 |
+
from pinder.core.index.system import PinderSystem
|
53 |
+
from pinder.core.loader.structure import Structure
|
54 |
+
from pinder.core.utils import constants as pc
|
55 |
+
from pinder.core.utils.log import setup_logger
|
56 |
+
from pinder.core.index.system import _align_monomers_with_mask
|
57 |
+
from pinder.core.loader.structure import Structure
|
58 |
+
|
59 |
+
import torch
|
60 |
+
import torch.nn as nn
|
61 |
+
import torch.nn.functional as F
|
62 |
+
from torch_geometric.nn import MessagePassing
|
63 |
+
from torch_geometric.nn import global_mean_pool
|
64 |
+
from torch.nn import Sequential, Linear, BatchNorm1d, ReLU
|
65 |
+
from torch_scatter import scatter
|
66 |
+
from torch.nn import Module
|
67 |
+
import time
|
68 |
+
from torch_geometric.nn import global_max_pool
|
69 |
+
import copy
|
70 |
+
import inspect
|
71 |
+
import warnings
|
72 |
+
from typing import Optional, Tuple, Union
|
73 |
+
|
74 |
+
import torch
|
75 |
+
from torch import Tensor
|
76 |
+
|
77 |
+
from torch_geometric.data import Data, Dataset, HeteroData
|
78 |
+
from torch_geometric.data.feature_store import FeatureStore
|
79 |
+
from torch_geometric.data.graph_store import GraphStore
|
80 |
+
from torch_geometric.loader import (
|
81 |
+
LinkLoader,
|
82 |
+
LinkNeighborLoader,
|
83 |
+
NeighborLoader,
|
84 |
+
NodeLoader,
|
85 |
+
)
|
86 |
+
from torch_geometric.loader.dataloader import DataLoader
|
87 |
+
from torch_geometric.loader.utils import get_edge_label_index, get_input_nodes
|
88 |
+
from torch_geometric.sampler import BaseSampler, NeighborSampler
|
89 |
+
from torch_geometric.typing import InputEdges, InputNodes
|
90 |
+
|
91 |
+
try:
|
92 |
+
from lightning.pytorch import LightningDataModule as PLLightningDataModule
|
93 |
+
no_pytorch_lightning = False
|
94 |
+
except (ImportError, ModuleNotFoundError):
|
95 |
+
PLLightningDataModule = object
|
96 |
+
no_pytorch_lightning = True
|
97 |
+
|
98 |
+
from lightning.pytorch.callbacks import ModelCheckpoint
|
99 |
+
from lightning.pytorch.loggers.tensorboard import TensorBoardLogger
|
100 |
+
from lightning.pytorch.callbacks.early_stopping import EarlyStopping
|
101 |
+
from torch_geometric.data.lightning.datamodule import LightningDataset
|
102 |
+
from pytorch_lightning.loggers.wandb import WandbLogger
|
103 |
+
def get_system(system_id: str) -> PinderSystem:
|
104 |
+
return PinderSystem(system_id)
|
105 |
+
from Bio import PDB
|
106 |
+
from Bio.PDB.PDBIO import PDBIO
|
107 |
+
|
108 |
+
log = setup_logger(__name__)
|
109 |
+
|
110 |
+
try:
|
111 |
+
from torch_cluster import knn_graph
|
112 |
+
|
113 |
+
torch_cluster_installed = True
|
114 |
+
except ImportError as e:
|
115 |
+
log.warning(
|
116 |
+
"torch-cluster is not installed!"
|
117 |
+
"Please install the appropriate library for your pytorch installation."
|
118 |
+
"See https://github.com/rusty1s/pytorch_cluster/issues/185 for background."
|
119 |
+
)
|
120 |
+
torch_cluster_installed = False
|
121 |
+
|
122 |
+
|
123 |
+
def structure2tensor(
|
124 |
+
atom_coordinates: NDArray[np.double] | None = None,
|
125 |
+
atom_types: NDArray[np.str_] | None = None,
|
126 |
+
element_types: NDArray[np.str_] | None = None,
|
127 |
+
residue_coordinates: NDArray[np.double] | None = None,
|
128 |
+
residue_ids: NDArray[np.int_] | None = None,
|
129 |
+
residue_types: NDArray[np.str_] | None = None,
|
130 |
+
chain_ids: NDArray[np.str_] | None = None,
|
131 |
+
dtype: torch.dtype = torch.float32,
|
132 |
+
) -> dict[str, torch.Tensor]:
|
133 |
+
property_dict = {}
|
134 |
+
if atom_types is not None:
|
135 |
+
unknown_name_idx = max(pc.ALL_ATOM_POSNS.values()) + 1
|
136 |
+
types_array_at = np.zeros((len(atom_types), 1))
|
137 |
+
for i, name in enumerate(atom_types):
|
138 |
+
types_array_at[i] = pc.ALL_ATOM_POSNS.get(name, unknown_name_idx)
|
139 |
+
property_dict["atom_types"] = torch.tensor(types_array_at).type(dtype)
|
140 |
+
if element_types is not None:
|
141 |
+
types_array_ele = np.zeros((len(element_types), 1))
|
142 |
+
for i, name in enumerate(element_types):
|
143 |
+
types_array_ele[i] = pc.ELE2NUM.get(name, pc.ELE2NUM["other"])
|
144 |
+
property_dict["element_types"] = torch.tensor(types_array_ele).type(dtype)
|
145 |
+
if residue_types is not None:
|
146 |
+
unknown_name_idx = max(pc.AA_TO_INDEX.values()) + 1
|
147 |
+
types_array_res = np.zeros((len(residue_types), 1))
|
148 |
+
for i, name in enumerate(residue_types):
|
149 |
+
types_array_res[i] = pc.AA_TO_INDEX.get(name, unknown_name_idx)
|
150 |
+
property_dict["residue_types"] = torch.tensor(types_array_res).type(dtype)
|
151 |
+
|
152 |
+
if atom_coordinates is not None:
|
153 |
+
property_dict["atom_coordinates"] = torch.tensor(atom_coordinates, dtype=dtype)
|
154 |
+
|
155 |
+
if residue_coordinates is not None:
|
156 |
+
property_dict["residue_coordinates"] = torch.tensor(
|
157 |
+
residue_coordinates, dtype=dtype
|
158 |
+
)
|
159 |
+
if residue_ids is not None:
|
160 |
+
property_dict["residue_ids"] = torch.tensor(residue_ids, dtype=dtype)
|
161 |
+
if chain_ids is not None:
|
162 |
+
property_dict["chain_ids"] = torch.zeros(len(chain_ids), dtype=dtype)
|
163 |
+
property_dict["chain_ids"][chain_ids == "L"] = 1
|
164 |
+
return property_dict
|
165 |
+
|
166 |
+
|
167 |
+
class NodeRepresentation(Enum):
|
168 |
+
Surface = "surface"
|
169 |
+
Atom = "atom"
|
170 |
+
Residue = "residue"
|
171 |
+
|
172 |
+
|
173 |
+
class PairedPDB(HeteroData): # type: ignore
|
174 |
+
@classmethod
|
175 |
+
def from_tuple_system(
|
176 |
+
cls,
|
177 |
+
|
178 |
+
tupal: tuple = (Structure , Structure , Structure),
|
179 |
+
|
180 |
+
add_edges: bool = True,
|
181 |
+
k: int = 10,
|
182 |
+
|
183 |
+
) -> PairedPDB:
|
184 |
+
return cls.from_structure_pair(
|
185 |
+
|
186 |
+
holo=tupal[0],
|
187 |
+
apo=tupal[1],
|
188 |
+
add_edges=add_edges,
|
189 |
+
k=k,
|
190 |
+
)
|
191 |
+
|
192 |
+
@classmethod
|
193 |
+
def from_structure_pair(
|
194 |
+
cls,
|
195 |
+
|
196 |
+
holo: Structure,
|
197 |
+
apo: Structure,
|
198 |
+
|
199 |
+
add_edges: bool = True,
|
200 |
+
k: int = 10,
|
201 |
+
) -> PairedPDB:
|
202 |
+
graph = cls()
|
203 |
+
holo_calpha = holo.filter("atom_name", mask=["CA"])
|
204 |
+
apo_calpha = apo.filter("atom_name", mask=["CA"])
|
205 |
+
r_h = (holo.dataframe['chain_id'] == 'R').sum()
|
206 |
+
r_a = (apo.dataframe['chain_id'] == 'R').sum()
|
207 |
+
|
208 |
+
holo_r_props = structure2tensor(
|
209 |
+
atom_coordinates=holo.coords[:r_h],
|
210 |
+
atom_types=holo.atom_array.atom_name[:r_h],
|
211 |
+
element_types=holo.atom_array.element[:r_h],
|
212 |
+
residue_coordinates=holo_calpha.coords[:r_h],
|
213 |
+
residue_types=holo_calpha.atom_array.res_name[:r_h],
|
214 |
+
residue_ids=holo_calpha.atom_array.res_id[:r_h],
|
215 |
+
)
|
216 |
+
holo_l_props = structure2tensor(
|
217 |
+
atom_coordinates=holo.coords[r_h:],
|
218 |
+
|
219 |
+
atom_types=holo.atom_array.atom_name[r_h:],
|
220 |
+
element_types=holo.atom_array.element[r_h:],
|
221 |
+
residue_coordinates=holo_calpha.coords[r_h:],
|
222 |
+
residue_types=holo_calpha.atom_array.res_name[r_h:],
|
223 |
+
residue_ids=holo_calpha.atom_array.res_id[r_h:],
|
224 |
+
)
|
225 |
+
apo_r_props = structure2tensor(
|
226 |
+
atom_coordinates=apo.coords[:r_a],
|
227 |
+
atom_types=apo.atom_array.atom_name[:r_a],
|
228 |
+
element_types=apo.atom_array.element[:r_a],
|
229 |
+
residue_coordinates=apo_calpha.coords[:r_a],
|
230 |
+
residue_types=apo_calpha.atom_array.res_name[:r_a],
|
231 |
+
residue_ids=apo_calpha.atom_array.res_id[:r_a],
|
232 |
+
)
|
233 |
+
apo_l_props = structure2tensor(
|
234 |
+
atom_coordinates=apo.coords[r_a:],
|
235 |
+
atom_types=apo.atom_array.atom_name[r_a:],
|
236 |
+
element_types=apo.atom_array.element[r_a:],
|
237 |
+
residue_coordinates=apo_calpha.coords[r_a:],
|
238 |
+
residue_types=apo_calpha.atom_array.res_name[r_a:],
|
239 |
+
residue_ids=apo_calpha.atom_array.res_id[r_a:],
|
240 |
+
)
|
241 |
+
|
242 |
+
|
243 |
+
|
244 |
+
graph["ligand"].x = apo_l_props["atom_types"]
|
245 |
+
graph["ligand"].pos = apo_l_props["atom_coordinates"]
|
246 |
+
graph["receptor"].x = apo_r_props["atom_types"]
|
247 |
+
graph["receptor"].pos = apo_r_props["atom_coordinates"]
|
248 |
+
graph["ligand"].y = holo_l_props["atom_coordinates"]
|
249 |
+
# graph["ligand"].pos = holo_l_props["atom_coordinates"]
|
250 |
+
graph["receptor"].y = holo_r_props["atom_coordinates"]
|
251 |
+
# graph["receptor"].pos = holo_r_props["atom_coordinates"]
|
252 |
+
if add_edges and torch_cluster_installed:
|
253 |
+
graph["ligand"].edge_index = knn_graph(
|
254 |
+
graph["ligand"].pos, k=k
|
255 |
+
)
|
256 |
+
graph["receptor"].edge_index = knn_graph(
|
257 |
+
graph["receptor"].pos, k=k
|
258 |
+
)
|
259 |
+
# graph["ligand"].edge_index = knn_graph(
|
260 |
+
# graph["ligand"].pos, k=k
|
261 |
+
# )
|
262 |
+
# graph["receptor"].edge_index = knn_graph(
|
263 |
+
# graph["receptor"].pos, k=k
|
264 |
+
# )
|
265 |
+
|
266 |
+
return graph
|
267 |
+
|
268 |
+
# To create dataset, we have used only PINDER datyaset with following steps as follows:
|
269 |
+
|
270 |
+
# log = setup_logger(__name__)
|
271 |
+
|
272 |
+
# try:
|
273 |
+
# from torch_cluster import knn_graph
|
274 |
+
|
275 |
+
# torch_cluster_installed = True
|
276 |
+
# except ImportError as e:
|
277 |
+
# log.warning(
|
278 |
+
# "torch-cluster is not installed!"
|
279 |
+
# "Please install the appropriate library for your pytorch installation."
|
280 |
+
# "See https://github.com/rusty1s/pytorch_cluster/issues/185 for background."
|
281 |
+
# )
|
282 |
+
# torch_cluster_installed = False
|
283 |
+
|
284 |
+
|
285 |
+
# def structure2tensor(
|
286 |
+
# atom_coordinates: NDArray[np.double] | None = None,
|
287 |
+
# atom_types: NDArray[np.str_] | None = None,
|
288 |
+
# element_types: NDArray[np.str_] | None = None,
|
289 |
+
# residue_coordinates: NDArray[np.double] | None = None,
|
290 |
+
# residue_ids: NDArray[np.int_] | None = None,
|
291 |
+
# residue_types: NDArray[np.str_] | None = None,
|
292 |
+
# chain_ids: NDArray[np.str_] | None = None,
|
293 |
+
# dtype: torch.dtype = torch.float32,
|
294 |
+
# ) -> dict[str, torch.Tensor]:
|
295 |
+
# property_dict = {}
|
296 |
+
# if atom_types is not None:
|
297 |
+
# unknown_name_idx = max(pc.ALL_ATOM_POSNS.values()) + 1
|
298 |
+
# types_array_at = np.zeros((len(atom_types), 1))
|
299 |
+
# for i, name in enumerate(atom_types):
|
300 |
+
# types_array_at[i] = pc.ALL_ATOM_POSNS.get(name, unknown_name_idx)
|
301 |
+
# property_dict["atom_types"] = torch.tensor(types_array_at).type(dtype)
|
302 |
+
# if element_types is not None:
|
303 |
+
# types_array_ele = np.zeros((len(element_types), 1))
|
304 |
+
# for i, name in enumerate(element_types):
|
305 |
+
# types_array_ele[i] = pc.ELE2NUM.get(name, pc.ELE2NUM["other"])
|
306 |
+
# property_dict["element_types"] = torch.tensor(types_array_ele).type(dtype)
|
307 |
+
# if residue_types is not None:
|
308 |
+
# unknown_name_idx = max(pc.AA_TO_INDEX.values()) + 1
|
309 |
+
# types_array_res = np.zeros((len(residue_types), 1))
|
310 |
+
# for i, name in enumerate(residue_types):
|
311 |
+
# types_array_res[i] = pc.AA_TO_INDEX.get(name, unknown_name_idx)
|
312 |
+
# property_dict["residue_types"] = torch.tensor(types_array_res).type(dtype)
|
313 |
+
|
314 |
+
# if atom_coordinates is not None:
|
315 |
+
# property_dict["atom_coordinates"] = torch.tensor(atom_coordinates, dtype=dtype)
|
316 |
+
|
317 |
+
# if residue_coordinates is not None:
|
318 |
+
# property_dict["residue_coordinates"] = torch.tensor(
|
319 |
+
# residue_coordinates, dtype=dtype
|
320 |
+
# )
|
321 |
+
# if residue_ids is not None:
|
322 |
+
# property_dict["residue_ids"] = torch.tensor(residue_ids, dtype=dtype)
|
323 |
+
# if chain_ids is not None:
|
324 |
+
# property_dict["chain_ids"] = torch.zeros(len(chain_ids), dtype=dtype)
|
325 |
+
# property_dict["chain_ids"][chain_ids == "L"] = 1
|
326 |
+
# return property_dict
|
327 |
+
|
328 |
+
|
329 |
+
# class NodeRepresentation(Enum):
|
330 |
+
# Surface = "surface"
|
331 |
+
# Atom = "atom"
|
332 |
+
# Residue = "residue"
|
333 |
+
|
334 |
+
|
335 |
+
# class PairedPDB(HeteroData): # type: ignore
|
336 |
+
# @classmethod
|
337 |
+
# def from_tuple_system(
|
338 |
+
# cls,
|
339 |
+
|
340 |
+
# tupal: tuple = (Structure , Structure , Structure),
|
341 |
+
|
342 |
+
# add_edges: bool = True,
|
343 |
+
# k: int = 10,
|
344 |
+
|
345 |
+
# ) -> PairedPDB:
|
346 |
+
# return cls.from_structure_pair(
|
347 |
+
|
348 |
+
# holo=tupal[0],
|
349 |
+
# apo=tupal[1],
|
350 |
+
# add_edges=add_edges,
|
351 |
+
# k=k,
|
352 |
+
# )
|
353 |
+
|
354 |
+
# @classmethod
|
355 |
+
# def from_structure_pair(
|
356 |
+
# cls,
|
357 |
+
|
358 |
+
# holo: Structure,
|
359 |
+
# apo: Structure,
|
360 |
+
|
361 |
+
# add_edges: bool = True,
|
362 |
+
# k: int = 10,
|
363 |
+
# ) -> PairedPDB:
|
364 |
+
# graph = cls()
|
365 |
+
# holo_calpha = holo.filter("atom_name", mask=["CA"])
|
366 |
+
# apo_calpha = apo.filter("atom_name", mask=["CA"])
|
367 |
+
# r_h = (holo.dataframe['chain_id'] == 'R').sum()
|
368 |
+
# r_a = (apo.dataframe['chain_id'] == 'R').sum()
|
369 |
+
|
370 |
+
# holo_r_props = structure2tensor(
|
371 |
+
# atom_coordinates=holo.coords[:r_h],
|
372 |
+
# atom_types=holo.atom_array.atom_name[:r_h],
|
373 |
+
# element_types=holo.atom_array.element[:r_h],
|
374 |
+
# residue_coordinates=holo_calpha.coords[:r_h],
|
375 |
+
# residue_types=holo_calpha.atom_array.res_name[:r_h],
|
376 |
+
# residue_ids=holo_calpha.atom_array.res_id[:r_h],
|
377 |
+
# )
|
378 |
+
# holo_l_props = structure2tensor(
|
379 |
+
# atom_coordinates=holo.coords[r_h:],
|
380 |
+
|
381 |
+
# atom_types=holo.atom_array.atom_name[r_h:],
|
382 |
+
# element_types=holo.atom_array.element[r_h:],
|
383 |
+
# residue_coordinates=holo_calpha.coords[r_h:],
|
384 |
+
# residue_types=holo_calpha.atom_array.res_name[r_h:],
|
385 |
+
# residue_ids=holo_calpha.atom_array.res_id[r_h:],
|
386 |
+
# )
|
387 |
+
# apo_r_props = structure2tensor(
|
388 |
+
# atom_coordinates=apo.coords[:r_a],
|
389 |
+
# atom_types=apo.atom_array.atom_name[:r_a],
|
390 |
+
# element_types=apo.atom_array.element[:r_a],
|
391 |
+
# residue_coordinates=apo_calpha.coords[:r_a],
|
392 |
+
# residue_types=apo_calpha.atom_array.res_name[:r_a],
|
393 |
+
# residue_ids=apo_calpha.atom_array.res_id[:r_a],
|
394 |
+
# )
|
395 |
+
# apo_l_props = structure2tensor(
|
396 |
+
# atom_coordinates=apo.coords[r_a:],
|
397 |
+
# atom_types=apo.atom_array.atom_name[r_a:],
|
398 |
+
# element_types=apo.atom_array.element[r_a:],
|
399 |
+
# residue_coordinates=apo_calpha.coords[r_a:],
|
400 |
+
# residue_types=apo_calpha.atom_array.res_name[r_a:],
|
401 |
+
# residue_ids=apo_calpha.atom_array.res_id[r_a:],
|
402 |
+
# )
|
403 |
+
|
404 |
+
|
405 |
+
|
406 |
+
# graph["ligand"].x = apo_l_props["atom_types"]
|
407 |
+
# graph["ligand"].pos = apo_l_props["atom_coordinates"]
|
408 |
+
# graph["receptor"].x = apo_r_props["atom_types"]
|
409 |
+
# graph["receptor"].pos = apo_r_props["atom_coordinates"]
|
410 |
+
# graph["ligand"].y = holo_l_props["atom_coordinates"]
|
411 |
+
# # graph["ligand"].pos = holo_l_props["atom_coordinates"]
|
412 |
+
# graph["receptor"].y = holo_r_props["atom_coordinates"]
|
413 |
+
# # graph["receptor"].pos = holo_r_props["atom_coordinates"]
|
414 |
+
# if add_edges and torch_cluster_installed:
|
415 |
+
# graph["ligand"].edge_index = knn_graph(
|
416 |
+
# graph["ligand"].pos, k=k
|
417 |
+
# )
|
418 |
+
# graph["receptor"].edge_index = knn_graph(
|
419 |
+
# graph["receptor"].pos, k=k
|
420 |
+
# )
|
421 |
+
# # graph["ligand"].edge_index = knn_graph(
|
422 |
+
# # graph["ligand"].pos, k=k
|
423 |
+
# # )
|
424 |
+
# # graph["receptor"].edge_index = knn_graph(
|
425 |
+
# # graph["receptor"].pos, k=k
|
426 |
+
# # )
|
427 |
+
|
428 |
+
# return graph
|
429 |
+
|
430 |
+
# index = get_index()
|
431 |
+
# # train = index[index.split == "train"].copy()
|
432 |
+
# # val = index[index.split == "val"].copy()
|
433 |
+
# # test = index[index.split == "test"].copy()
|
434 |
+
# # train_filtered = train[(train['apo_R'] == True) & (train['apo_L'] == True)].copy()
|
435 |
+
# # val_filtered = val[(val['apo_R'] == True) & (val['apo_L'] == True)].copy()
|
436 |
+
# # test_filtered = test[(test['apo_R'] == True) & (test['apo_L'] == True)].copy()
|
437 |
+
|
438 |
+
# # train_apo = [get_system(train_filtered.id.iloc[i]).create_masked_bound_unbound_complexes(
|
439 |
+
# # monomer_types=["apo"], renumber_residues=True
|
440 |
+
# # ) for i in range(0, 10000)]
|
441 |
+
|
442 |
+
# # train_new_apo11 = [get_system(train_filtered.id.iloc[i]).create_masked_bound_unbound_complexes(
|
443 |
+
# # monomer_types=["apo"], renumber_residues=True
|
444 |
+
# # ) for i in range(10000,10908)]
|
445 |
+
|
446 |
+
# # train_new_apo12 = [get_system(train_filtered.id.iloc[i]).create_masked_bound_unbound_complexes(
|
447 |
+
# # # monomer_types=["apo"], renumber_residues=True
|
448 |
+
# # ) for i in range(10908,11816)]
|
449 |
+
|
450 |
+
# # val_new_apo1 = [get_system(val_filtered.id.iloc[i]).create_masked_bound_unbound_complexes(
|
451 |
+
# # monomer_types=["apo"], renumber_residues=True
|
452 |
+
# # ) for i in range(0,342)]
|
453 |
+
|
454 |
+
# # test_new_apo1 = [get_system(test_filtered.id.iloc[i]).create_masked_bound_unbound_complexes(
|
455 |
+
# # monomer_types=["apo"], renumber_residues=True
|
456 |
+
# # ) for i in range(0,342)]
|
457 |
+
|
458 |
+
# # val_apo = val_new_apo1 + train_new_apo11
|
459 |
+
# # test_apo = test_new_apo1 + train_new_apo12
|
460 |
+
|
461 |
+
# import pickle
|
462 |
+
# # with open("train_apo.pkl", "wb") as file:
|
463 |
+
# # pickle.dump(train_apo, file)
|
464 |
+
|
465 |
+
# # with open("val_apo.pkl", "wb") as file:
|
466 |
+
# # pickle.dump(val_apo, file)
|
467 |
+
|
468 |
+
# # with open("test_apo.pkl", "wb") as file:
|
469 |
+
# # pickle.dump(test_apo, file)
|
470 |
+
# with open("train_apo.pkl", "rb") as file:
|
471 |
+
# train_apo = pickle.load(file)
|
472 |
+
|
473 |
+
# with open("val_apo.pkl", "rb") as file:
|
474 |
+
# val_apo = pickle.load(file)
|
475 |
+
|
476 |
+
# with open("test_apo.pkl", "rb") as file:
|
477 |
+
# test_apo = pickle.load(file)
|
478 |
+
|
479 |
+
# # # %%
|
480 |
+
# train_geo = [PairedPDB.from_tuple_system(train_apo[i]) for i in range(0,len(train_apo))]
|
481 |
+
# val_geo = [PairedPDB.from_tuple_system(val_apo[i]) for i in range(0,len(val_apo))]
|
482 |
+
# test_geo = [PairedPDB.from_tuple_system(test_apo[i]) for i in range(0,len(test_apo))]
|
483 |
+
# # # %%
|
484 |
+
# # Train= []
|
485 |
+
# # for i in range(0,len(train_geo)):
|
486 |
+
# # data = HeteroData()
|
487 |
+
# # data["ligand"].x = train_geo[i]["ligand"].x
|
488 |
+
# # data['ligand'].y = train_geo[i]["ligand"].y
|
489 |
+
# # data["ligand"].pos = train_geo[i]["ligand"].pos
|
490 |
+
# # data["ligand","ligand"].edge_index = train_geo[i]["ligand"]
|
491 |
+
# # data["receptor"].x = train_geo[i]["receptor"].x
|
492 |
+
# # data['receptor'].y = train_geo[i]["receptor"].y
|
493 |
+
# # data["receptor"].pos = train_geo[i]["receptor"].pos
|
494 |
+
# # data["receptor","receptor"].edge_index = train_geo[i]["receptor"]
|
495 |
+
# # #torch.save(data, f"./data/processed/train_sample_{i}.pt")
|
496 |
+
# # Train.append(data)
|
497 |
+
|
498 |
+
# from torch_geometric.data import HeteroData
|
499 |
+
# import torch_sparse
|
500 |
+
# from torch_geometric.edge_index import to_sparse_tensor
|
501 |
+
# import torch
|
502 |
+
|
503 |
+
# # Example of converting edge indices to SparseTensor and storing them in HeteroData
|
504 |
+
|
505 |
+
# Train1 = []
|
506 |
+
# for i in range(len(train_geo)):
|
507 |
+
# data = HeteroData()
|
508 |
+
# # Define ligand node features
|
509 |
+
# data["ligand"].x = train_geo[i]["ligand"].x
|
510 |
+
# data["ligand"].y = train_geo[i]["ligand"].y
|
511 |
+
# data["ligand"].pos = train_geo[i]["ligand"].pos
|
512 |
+
# # Convert ligand edge index to SparseTensor
|
513 |
+
# ligand_edge_index = train_geo[i]["ligand"]["edge_index"]
|
514 |
+
# data["ligand", "ligand"].edge_index = to_sparse_tensor(ligand_edge_index, sparse_sizes=(train_geo[i]["ligand"].num_nodes,)*2)
|
515 |
+
|
516 |
+
# # Define receptor node features
|
517 |
+
# data["receptor"].x = train_geo[i]["receptor"].x
|
518 |
+
# data["receptor"].y = train_geo[i]["receptor"].y
|
519 |
+
# data["receptor"].pos = train_geo[i]["receptor"].pos
|
520 |
+
# # Convert receptor edge index to SparseTensor
|
521 |
+
# receptor_edge_index = train_geo[i]["receptor"]["edge_index"]
|
522 |
+
# data["receptor", "receptor"].edge_index = to_sparse_tensor(receptor_edge_index, sparse_sizes=(train_geo[i]["receptor"].num_nodes,)*2)
|
523 |
+
|
524 |
+
# Train1.append(data)
|
525 |
+
|
526 |
+
|
527 |
+
# # # %%
|
528 |
+
# # Val= []
|
529 |
+
# # for i in range(0,len(val_geo)):
|
530 |
+
# # data = HeteroData()
|
531 |
+
# # data["ligand"].x = val_geo[i]["ligand"].x
|
532 |
+
# # data['ligand'].y = val_geo[i]["ligand"].y
|
533 |
+
# # data["ligand"].pos = val_geo[i]["ligand"].pos
|
534 |
+
# # data["ligand","ligand"].edge_index = val_geo[i]["ligand"]
|
535 |
+
# # data["receptor"].x = val_geo[i]["receptor"].x
|
536 |
+
# # data['receptor'].y = val_geo[i]["receptor"].y
|
537 |
+
# # data["receptor"].pos = val_geo[i]["receptor"].pos
|
538 |
+
# # data["receptor","receptor"].edge_index = val_geo[i]["receptor"]
|
539 |
+
# # #torch.save(data, f"./data/processed/val_sample_{i}.pt")
|
540 |
+
# # Val.append(data)
|
541 |
+
# Val1 = []
|
542 |
+
# for i in range(len(val_geo)):
|
543 |
+
# data = HeteroData()
|
544 |
+
# # Define ligand node features
|
545 |
+
# data["ligand"].x = val_geo[i]["ligand"].x
|
546 |
+
# data["ligand"].y = val_geo[i]["ligand"].y
|
547 |
+
# data["ligand"].pos = val_geo[i]["ligand"].pos
|
548 |
+
# # Convert ligand edge index to SparseTensor
|
549 |
+
# ligand_edge_index = val_geo[i]["ligand"]["edge_index"]
|
550 |
+
# data["ligand", "ligand"].edge_index = to_sparse_tensor(ligand_edge_index, sparse_sizes=(val_geo[i]["ligand"].num_nodes,)*2)
|
551 |
+
|
552 |
+
# # Define receptor node features
|
553 |
+
# data["receptor"].x = val_geo[i]["receptor"].x
|
554 |
+
# data["receptor"].y = val_geo[i]["receptor"].y
|
555 |
+
# data["receptor"].pos = val_geo[i]["receptor"].pos
|
556 |
+
# # Convert receptor edge index to SparseTensor
|
557 |
+
# receptor_edge_index = val_geo[i]["receptor"]["edge_index"]
|
558 |
+
# data["receptor", "receptor"].edge_index = to_sparse_tensor(receptor_edge_index, sparse_sizes=(val_geo[i]["receptor"].num_nodes,)*2)
|
559 |
+
|
560 |
+
# Val1.append(data)
|
561 |
+
# # # %%
|
562 |
+
# # Test= []
|
563 |
+
# # for i in range(0,len(test_geo)):
|
564 |
+
# # data = HeteroData()
|
565 |
+
# # data["ligand"].x = test_geo[i]["ligand"].x
|
566 |
+
# # data['ligand'].y = test_geo[i]["ligand"].y
|
567 |
+
# # data["ligand"].pos = test_geo[i]["ligand"].pos
|
568 |
+
# # data["ligand","ligand"].edge_index = test_geo[i]["ligand"]
|
569 |
+
# # data["receptor"].x = test_geo[i]["receptor"].x
|
570 |
+
# # data['receptor'].y = test_geo[i]["receptor"].y
|
571 |
+
# # data["receptor"].pos = test_geo[i]["receptor"].pos
|
572 |
+
# # data["receptor","receptor"].edge_index = test_geo[i]["receptor"]
|
573 |
+
# # #torch.save(data, f"./data/processed/test_sample_{i}.pt")
|
574 |
+
# # Test.append(data)
|
575 |
+
# Test1 = []
|
576 |
+
# for i in range(len(test_geo)):
|
577 |
+
# data = HeteroData()
|
578 |
+
# # Define ligand node features
|
579 |
+
# data["ligand"].x = test_geo[i]["ligand"].x
|
580 |
+
# data["ligand"].y = test_geo[i]["ligand"].y
|
581 |
+
# data["ligand"].pos = test_geo[i]["ligand"].pos
|
582 |
+
# # Convert ligand edge index to SparseTensor
|
583 |
+
# ligand_edge_index = test_geo[i]["ligand"]["edge_index"]
|
584 |
+
# data["ligand", "ligand"].edge_index = to_sparse_tensor(ligand_edge_index, sparse_sizes=(test_geo[i]["ligand"].num_nodes,)*2)
|
585 |
+
|
586 |
+
# # Define receptor node features
|
587 |
+
# data["receptor"].x = test_geo[i]["receptor"].x
|
588 |
+
# data["receptor"].y = test_geo[i]["receptor"].y
|
589 |
+
# data["receptor"].pos = test_geo[i]["receptor"].pos
|
590 |
+
# # Convert receptor edge index to SparseTensor
|
591 |
+
# receptor_edge_index = test_geo[i]["receptor"]["edge_index"]
|
592 |
+
# data["receptor", "receptor"].edge_index = to_sparse_tensor(receptor_edge_index, sparse_sizes=(test_geo[i]["receptor"].num_nodes,)*2)
|
593 |
+
|
594 |
+
# Test1.append(data)
|
595 |
+
# # with open("Train.pkl", "wb") as file:
|
596 |
+
# # pickle.dump(Train, file)
|
597 |
+
|
598 |
+
# # with open("Val.pkl", "wb") as file:
|
599 |
+
# # pickle.dump(Val, file)
|
600 |
+
|
601 |
+
# # with open("Test.pkl", "wb") as file:
|
602 |
+
# # pickle.dump(Test, file)
|
603 |
+
|
604 |
+
# # with open("Train1.pkl", "rb") as file:
|
605 |
+
# # Train= pickle.load(file)
|
606 |
+
|
607 |
+
# # with open("Val.pkl", "rb") as file:
|
608 |
+
# # Val = pickle.load(file)
|
609 |
+
|
610 |
+
# # with open("Test.pkl", "rb") as file:
|
611 |
+
# # Test = pickle.load(file)
|