File size: 7,522 Bytes
2596438 fb83fd2 2596438 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 |
from __future__ import annotations
from pathlib import Path
import time
from biotite.application.autodock import VinaApp
import gradio as gr
from gradio_molecule3d import Molecule3D
from gradio_molecule2d import molecule2d
import numpy as np
from rdkit import Chem
from rdkit.Chem import AllChem
import pandas as pd
from biotite.structure import centroid, from_template
from biotite.structure.io import load_structure
from biotite.structure.io.mol import MOLFile, SDFile
from biotite.structure.io.pdb import PDBFile
from plinder.eval.docking.write_scores import evaluate
EVAL_METRICS = ["system", "LDDT-PLI", "LDDT-LP", "BISY-RMSD"]
def vina(
ligand, receptor, pocket_center, output_folder: Path, size=10, max_num_poses=5
):
app = VinaApp(
ligand,
receptor,
center=pocket_center,
size=[size, size, size],
)
app.set_max_number_of_models(max_num_poses)
app.start()
app.join()
docked_ligand = from_template(ligand, app.get_ligand_coord())
docked_ligand = docked_ligand[..., ~np.isnan(docked_ligand.coord[0]).any(axis=-1)]
output_files = []
for i in range(max_num_poses):
sdf_file = MOLFile()
sdf_file.set_structure(docked_ligand[i])
output_file = output_folder / f"docked_ligand_{i}.sdf"
sdf_file.write(output_file)
output_files.append(output_file)
return output_files
def predict(
input_sequence: str,
input_ligand: str,
input_msa: gr.File | None = None,
input_protein: gr.File | None = None,
max_num_poses: int = 1,
):
"""
Main prediction function that calls ligsite and smina
Parameters
----------
input_sequence: str
monomer sequence
input_ligand: str
ligand as SMILES string
input_msa: gradio.File | None
Gradio file object to MSA a3m file
input_protein: gradio.File | None
Gradio file object to monomer protein structure as CIF file
max_num_poses: int
Number of poses to generate
Returns
-------
output_structures: tuple
(output_protein, output_ligand_sdf)
run_time: float
run time of the program
"""
start_time = time.time()
if input_protein is None:
raise gr.Error("need input_protein")
print(input_protein)
ligand_file = Path(input_protein).parent / "ligand.sdf"
print(ligand_file)
conformer = Chem.AddHs(Chem.MolFromSmiles(input_ligand))
AllChem.EmbedMolecule(conformer)
AllChem.MMFFOptimizeMolecule(conformer)
Chem.SDWriter(ligand_file).write(conformer)
ligand = SDFile.read(ligand_file).record.get_structure()
receptor = load_structure(input_protein, include_bonds=True)
docking_poses = vina(
ligand,
receptor,
centroid(receptor),
Path(input_protein).parent,
max_num_poses=max_num_poses,
)
end_time = time.time()
run_time = end_time - start_time
pdb_file = PDBFile()
pdb_file.set_structure(receptor)
output_pdb = Path(input_protein).parent / "receptor.pdb"
pdb_file.write(output_pdb)
return [str(output_pdb), str(docking_poses[0])], run_time
def get_metrics(
system_id: str,
receptor_file: Path,
ligand_file: Path,
flexible: bool = True,
posebusters: bool = True,
) -> tuple[pd.DataFrame, float]:
start_time = time.time()
metrics = pd.DataFrame(
[
evaluate(
model_system_id=system_id,
reference_system_id=system_id,
receptor_file=receptor_file,
ligand_file_list=[Path(ligand_file)],
flexible=flexible,
posebusters=posebusters,
posebusters_full=False,
).get("LIG_0", {})
]
)
if posebusters:
metrics["posebusters"] = metrics[
[col for col in metrics.columns if col.startswith("posebusters_")]
].sum(axis=1)
metrics["posebusters_valid"] = metrics[
[col for col in metrics.columns if col.startswith("posebusters_")]
].sum(axis=1) == 20
columns = ["reference", "lddt_pli_ave", "lddt_lp_ave", "bisy_rmsd_ave"]
if flexible:
columns.extend(["lddt", "bb_lddt"])
if posebusters:
columns.extend([col for col in metrics.columns if col.startswith("posebusters")])
metrics = metrics[columns].copy()
mapping = {
"lddt_pli_ave": "LDDT-PLI",
"lddt_lp_ave": "LDDT-LP",
"bisy_rmsd_ave": "BISY-RMSD",
"reference": "system",
}
if flexible:
mapping["lddt"] = "LDDT"
mapping["bb_lddt"] = "Backbone LDDT"
if posebusters:
mapping["posebusters"] = "PoseBusters #checks"
mapping["posebusters_valid"] = "PoseBusters valid"
metrics.rename(
columns=mapping,
inplace=True,
)
end_time = time.time()
run_time = end_time - start_time
return metrics, run_time
with gr.Blocks() as app:
with gr.Tab("🧬 PINDER evaluation template"):
with gr.Row():
with gr.Column():
input_system_id_pinder = gr.Textbox(label="PINDER system ID")
input_receptor_file_pinder = gr.File(label="Receptor file")
input_ligand_file_pinder = gr.File(label="Ligand file")
methodname_pinder = gr.Textbox(label="Name of your method in the format mlsb/spacename")
store_pinder = gr.Checkbox(label="Store on huggingface for leaderboard", value=False)
eval_btn_pinder = gr.Button("Run Evaluation")
with gr.Tab("⚖️ PLINDER evaluation template"):
with gr.Row():
with gr.Column():
input_system_id = gr.Textbox(label="PLINDER system ID")
input_receptor_file = gr.File(label="Receptor file (CIF)")
input_ligand_file = gr.File(label="Ligand file (SDF)")
flexible = gr.Checkbox(label="Flexible docking", value=True)
posebusters = gr.Checkbox(label="PoseBusters", value=True)
methodname = gr.Textbox(label="Name of your method in the format mlsb/spacename")
store = gr.Checkbox(label="Store on huggingface for leaderboard", value=False)
eval_btn = gr.Button("Run Evaluation")
gr.Examples(
[
[
"4neh__1__1.B__1.H",
"input_protein_test.cif",
"input_ligand_test.sdf",
True,
True,
],
],
[input_system_id, input_receptor_file, input_ligand_file, flexible, posebusters, methodname, store],
)
eval_run_time = gr.Textbox(label="Evaluation runtime")
metric_table = gr.DataFrame(
pd.DataFrame([], columns=EVAL_METRICS), label="Evaluation metrics"
)
metric_table_pinder = gr.DataFrame(
pd.DataFrame([], columns=EVAL_METRICS_PINDER), label="Evaluation metrics"
)
eval_btn.click(
get_metrics,
inputs=[input_system_id, input_receptor_file, input_ligand_file, flexible, posebusters],
outputs=[metric_table, eval_run_time],
)
eval_btn_pinder.click(
get_metrics_pinder,
inputs=[input_system_id_pinder, input_receptor_file_pinder, input_ligand_file_pinder, methodname_pinder, store_pinder],
outputs=[metric_table_pinder, eval_run_time],
)
app.launch() |