File size: 9,199 Bytes
d766458 | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 | # Copyright 2025 ByteDance and/or its affiliates.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import json
import os
import subprocess
import tempfile
from typing import Any, Dict, List
from pxdbench.globals import MPNN_CKPT_PATH, _require
class VanillaMPNNPredictor:
def __init__(self, cfg, device_id: int = 0, seed: int = None, verbose=True):
self.cfg = cfg
self.device_id = device_id
self.seed = seed
self.verbose = verbose
_require(
os.path.join(
MPNN_CKPT_PATH[self.cfg["model_type"]], self.cfg["model_name"] + ".pt"
)
)
dir_name = os.path.dirname(__file__)
self.prepare_script_path = os.path.join(dir_name, "parse_multiple_chains.py")
self.run_script_path = os.path.join(dir_name, "protein_mpnn_run.py")
self.env = os.environ.copy()
self.env["CUDA_VISIBLE_DEVICES"] = str(device_id)
def prepare_jsonl(self, input_dir: str, pdb_names: list[str]):
"""
Prepare a JSONL file for ProteinMPNN input by parsing multiple chains from PDB files.
Args:
input_dir (str): Directory containing input PDB files.
pdb_names (list[str]): List of PDB base names (without '.pdb') to process.
Returns:
str: Path to the output JSONL file.
"""
with tempfile.NamedTemporaryFile(suffix=".jsonl", delete=False) as f:
output_path = f.name
output_path = os.path.join(input_dir, "parsed_pdbs.jsonl")
cmd = [
"python3",
"-u",
self.prepare_script_path,
"--input_path",
input_dir,
"--output",
output_path,
]
try:
process = subprocess.Popen(
cmd,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE,
text=True,
env=self.env,
)
while True:
output = process.stdout.readline()
if output == "" and process.poll() is not None:
break
if output:
print(output.strip())
while True:
error = process.stderr.readline()
if error == "" and process.poll() is not None:
break
if error:
print(error.strip())
returncode = process.wait()
print(f"Run subprocess success: {returncode}")
except Exception as e:
print(f"Run subprocess fail: {str(e)}")
raise
result = []
with open(output_path, "r") as infile:
for line in infile:
entry = json.loads(line)
if entry.get("name") in pdb_names:
result.append(entry)
with open(output_path, "w") as f:
for entry in result:
f.write(json.dumps(entry) + "\n")
return output_path
def run_mpnn(self, jsonl_path: str, pdb_names: list[str], num_seqs: int):
"""
Run ProteinMPNN to design sequences for the given PDB names.
Args:
jsonl_path (str): Path to the input JSONL file containing parsed PDB data.
pdb_names (list[str]): List of PDB base names (without '.pdb') to process.
num_seqs (int): Number of sequences to generate per PDB.
Returns:
None
"""
model_type = self.cfg["model_type"]
model_name = self.cfg["model_name"]
assert model_type in ["ca", "bb", "soluable"]
path_to_model_weights = MPNN_CKPT_PATH[model_type]
output_dir = os.path.dirname(jsonl_path)
cmd = [
"python3",
"-u",
self.run_script_path,
"--jsonl_path",
jsonl_path,
"--out_folder",
output_dir,
"--num_seq_per_target",
str(num_seqs),
"--sampling_temp",
self.cfg["temperature"],
"--batch_size",
str(num_seqs),
"--model_name",
model_name,
"--path_to_model_weights",
path_to_model_weights,
]
if model_type == "ca":
cmd.extend(["--ca_only"])
elif model_type == "soluble":
cmd.extend(["--use_soluble_model"])
if self.seed is not None:
cmd.extend(["--seed", str(self.seed)])
print("Run Protein MPNN with %s" % (" ".join(cmd)))
try:
process = subprocess.Popen(
cmd,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE,
text=True,
env=self.env,
)
while True:
output = process.stdout.readline()
if output == "" and process.poll() is not None:
break
if output:
print(output.strip())
while True:
error = process.stderr.readline()
if error == "" and process.poll() is not None:
break
if error:
print(error.strip())
returncode = process.wait()
print(f"Run subprocess success: {returncode}")
except Exception as e:
print(f"Run subprocess fail: {str(e)}")
raise
# postprocess
seq_dir = os.path.join(output_dir, "seqs")
result = []
for name in pdb_names:
fasta_file = name + ".fa"
fasta_path = os.path.join(seq_dir, fasta_file)
if not os.path.exists(fasta_path):
continue
with open(fasta_path, "r") as f:
lines = f.readlines()
for i, line_no in enumerate(
range(4, int(num_seqs) * 2 + 3, 2)
): # lines 4,6,...,18 (1-based)
if line_no <= len(lines):
sequence = lines[line_no - 1].strip()
result.append({"name": name, "seq_idx": i, "sequence": sequence})
return result
def design_monomer(
self, pdb_dir: str, pdb_names: List[str], num_samples: int
) -> List[Dict]:
"""
Design sequences for monomer proteins using ProteinMPNN.
Args:
pdb_dir (str): Directory containing input PDB files.
pdb_names (list[str]): List of PDB base names (without '.pdb') to process.
num_samples (int): Number of sequences to generate per PDB.
Returns:
list[dict]: List of design results with keys 'name' (PDB name), 'seq_idx' (sequence index),
and 'sequence' (designed amino acid sequence).
"""
# input_data = {
# "pdb_dir": pdb_dir,
# "pdb_names": pdb_names,
# "num_samples": num_samples,
# "mpnn_cfg": self.cfg.to_dict(),
# "design_type": "monomer",
# }
jsonl_path = self.prepare_jsonl(pdb_dir, pdb_names)
output = self.run_mpnn(jsonl_path, pdb_names, num_samples)
os.unlink(jsonl_path)
return output
def design_binder(
self,
pdb_dir: str,
pdb_names: List[str],
num_samples: int,
binder_chains: List[str],
cond_chains: List[str],
) -> List[Dict]:
"""
Design sequences for binder proteins using ProteinMPNN.
Args:
pdb_dir (str): Directory containing input PDB files.
pdb_names (list[str]): List of PDB base names (without '.pdb') to process.
num_samples (int): Number of sequences to generate per PDB.
binder_chains (list[str]): List of binder chain IDs.
cond_chains (list[str]): List of conditional chain IDs.
Returns:
list[dict]: List of design results with keys 'name' (PDB name), 'seq_idx' (sequence index),
and 'sequence' (designed amino acid sequence).
"""
input_data = {
"pdb_dir": pdb_dir,
"pdb_names": pdb_names,
"num_samples": num_samples,
"binder_chains": binder_chains,
"cond_chains": cond_chains,
"mpnn_cfg": self.cfg.to_dict(),
"design_type": "binder",
}
output = self.run(input_data)
return output
|