2025 - AIDD - python的autodock vina 批量分子对接改进版本2.0-全自动对接,完全全自动对接

2025 - AIDD - python的autodock vina 批量分子对接改进版本2.0-全自动对接,完全全自动对接

python 复制代码
import warnings
from pathlib import Path
import subprocess
from itertools import product
import numpy as np
import pandas as pd
from MDAnalysis import Universe
from openbabel import pybel
import os
import requests  # 用于下载文件

# 清楚 warning 信息

# Filter warnings
warnings.filterwarnings("ignore")

# 设置工作目录
HERE = Path(os.getcwd())
DATA = HERE / 'data'
DATA.mkdir(parents=True, exist_ok=True)

# CSV 文件路径
RESULT_CSV = DATA / 'result_dock.csv'

# 初始化 CSV 文件
if not RESULT_CSV.exists():
    pd.DataFrame(columns=['pdb', 'ligand_resnames', 'smiles', 'dock_info']).to_csv(RESULT_CSV, index=False)


class Structure(Universe):
    """Core object to load structures with."""

    @classmethod
    def from_string(cls, pdb_path):
        """Load a structure from a local PDB file."""
        return cls(pdb_path)


def download_pdb(pdb_id, save_path):
    """从 RCSB PDB 数据库下载 PDB 文件."""
    url = f"https://files.rcsb.org/download/{pdb_id}.pdb"
    response = requests.get(url)
    if response.status_code == 200:
        with open(save_path, "wb") as file:
            file.write(response.content)
        print(f"{pdb_id} 下载成功.")
    else:
        raise ValueError(f"无法下载 {pdb_id}: {response.status_code}")


# 提取蛋白质并保存为 PDB 格式
def extract_protein_to_pdb(structure, protein_path):
    """Extract protein from structure and save it as PDB."""
    protein = structure.select_atoms("protein")
    protein.write(str(protein_path))


# 自动检测所有配体残基
def find_all_ligand_resnames(structure):
    """返回所有非蛋白质和非水分子的残基名称列表"""
    ligand_atoms = structure.select_atoms("not protein and not resname HOH")
    return list(set(ligand_atoms.resnames))


def pdb_to_pdbqt(pdb_path, pdbqt_path, pH=7.4):
    """Convert a PDB file to a PDBQT file."""
    molecule = list(pybel.readfile("pdb", str(pdb_path)))[0]
    molecule.OBMol.CorrectForPH(pH)
    molecule.addh()
    for atom in molecule.atoms:
        atom.OBAtom.GetPartialCharge()
    molecule.write("pdbqt", str(pdbqt_path), overwrite=True)


def smiles_to_pdbqt(smiles, pdbqt_path, pH=7.4):
    """Convert a SMILES string to a PDBQT file."""
    molecule = pybel.readstring("smi", smiles)
    molecule.OBMol.CorrectForPH(pH)
    molecule.addh()
    molecule.make3D(forcefield="mmff94s", steps=10000)
    for atom in molecule.atoms:
        atom.OBAtom.GetPartialCharge()
    molecule.write("pdbqt", str(pdbqt_path), overwrite=True)


def run_smina(ligand_path, protein_path, out_path, pocket_center, pocket_size):
    """Perform docking with Smina."""
    output_text = subprocess.check_output([
        "smina",
        "--receptor", str(protein_path),
        "--ligand", str(ligand_path),
        "--out", str(out_path),
        "--center_x", str(pocket_center[0]),
        "--center_y", str(pocket_center[1]),
        "--center_z", str(pocket_center[2]),
        "--size_x", str(pocket_size[0]),
        "--size_y", str(pocket_size[1]),
        "--size_z", str(pocket_size[2])
    ])
    return output_text.decode("utf-8")


def parse_dock_info(dock_output):
    """解析 Smina 输出中的对接信息"""
    dock_lines = dock_output.splitlines()
    dock_data = []
    capture = False
    for line in dock_lines:
        if "mode" in line and "affinity" in line:  # 开始捕获表格数据
            capture = True
            continue
        if capture:
            if line.strip() == "" or "Refine time" in line or "Loop time" in line:
                break
            dock_data.append(line.strip())
    return "\n".join(dock_data)


def split_sdf_file(sdf_path):
    """Split an SDF file into separate files for each molecule."""
    sdf_path = Path(sdf_path)
    stem = sdf_path.stem
    parent = sdf_path.parent
    molecules = pybel.readfile("sdf", str(sdf_path))
    for i, molecule in enumerate(molecules, 1):
        molecule.write("sdf", str(parent / f"{stem}_{i}.sdf"), overwrite=True)



pdb_list = pd.read_csv('data/pdb.csv')['pdb_id']
# 读取 data 里面的 pic50_greater_equal_9.0.csv 文件,获取其中的 SMILES,前面的 20 个
smiles_list = pd.read_csv('data/pic50_greater_equal_9.0.csv')['smiles'][:20]

# 遍历所有 PDB 和 SMILES 的笛卡尔积
for pdb_id, smiles in product(pdb_list, smiles_list):
    # 创建每个 PDB 的独立目录
    pdb_dir = DATA / f"data_{pdb_id}"
    pdb_dir.mkdir(parents=True, exist_ok=True)

    # 定义 PDB 文件路径
    pdb_path = pdb_dir / f"{pdb_id}.pdb"

    # 如果 PDB 文件不存在,则从 RCSB 下载
    if not pdb_path.exists():
        print(f"{pdb_id} 文件不存在,正在下载...")
        download_pdb(pdb_id, pdb_path)

    # 加载结构
    structure = Structure.from_string(pdb_path)

    # 提取并保存蛋白质部分
    protein_path = pdb_dir / "protein.pdb"
    extract_protein_to_pdb(structure, protein_path)

    # 转换蛋白质为 PDBQT 格式
    protein_pdbqt_path = pdb_dir / "protein.pdbqt"
    pdb_to_pdbqt(protein_path, protein_pdbqt_path)

    # 检测所有 ligand
    all_ligands = find_all_ligand_resnames(structure)

    # all_ligands[0]变为shape为(1,)的数组,方便后续循环
    # all_ligands = ["VX6"]
    print(f"{pdb_id} - Detected Ligands: {all_ligands}")

    # 对所有 ligand 进行处理
    for ligand_resname in all_ligands:
        # 创建 ligand-specific 子目录
        ligand_dir = pdb_dir / f"ligand_{ligand_resname}"
        ligand_dir.mkdir(parents=True, exist_ok=True)

        # 自动检测口袋中心和大小
        ligand = structure.select_atoms(f"resname {ligand_resname}")
        print(f"Processing {pdb_id} {smiles} {ligand_resname}")
        pocket_center = (ligand.positions.max(axis=0) + ligand.positions.min(axis=0)) / 2
        pocket_size = ligand.positions.max(axis=0) - ligand.positions.min(axis=0) + 5

        # 创建针对 SMILES 的子目录
        smiles_hash = smiles.replace("(", "").replace(")", "").replace("=", "").replace("-", "").replace("/","").replace("\\", "").replace(".", "").replace(",", "").replace(":", "")
        smiles_dir = ligand_dir / f"smiles_{smiles_hash}"
        smiles_dir.mkdir(parents=True, exist_ok=True)

        # 定义文件路径
        ligand_path = smiles_dir / "ligand.pdbqt"
        docking_out_path = smiles_dir / "ligand_docking.sdf"
        log_path = smiles_dir / "docking_log.txt"

        # 转换配体为 PDBQT 格式
        smiles_to_pdbqt(smiles, ligand_path)

        # 运行对接并保存日志
        docking_output = run_smina(ligand_path, protein_pdbqt_path, docking_out_path, pocket_center, pocket_size)
        with open(log_path, "w") as log_file:
            log_file.write(f"PDB: {pdb_id}\nSMILES: {smiles}\nLigand Resname: {ligand_resname}\n")
            log_file.write(f"Pocket Center: {pocket_center}\nPocket Size: {pocket_size}\n\nDocking Output:\n")
            log_file.write(docking_output)

        # 拆分 SDF 文件
        split_sdf_file(docking_out_path)

        # 提取对接结果并更新到 CSV
        dock_info = parse_dock_info(docking_output)
        new_row = {'pdb': pdb_id, 'ligand_resnames': ligand_resname, 'smiles': smiles, 'dock_info': dock_info}
        result_df = pd.read_csv(RESULT_CSV)
        result_df = pd.concat([result_df, pd.DataFrame([new_row])], ignore_index=True)
        # 保存更新的结果到 CSV 文件
        result_df.to_csv(RESULT_CSV, index=False)

print("Docking workflow completed successfully!")
相关推荐
AI蜗牛之家1 小时前
Qwen系列之Qwen3解读:最强开源模型的细节拆解
人工智能·python
C++ 老炮儿的技术栈1 小时前
UDP 与 TCP 的区别是什么?
开发语言·c++·windows·算法·visual studio
wgslucky2 小时前
Dubbo报错:module java.base does not “opens java.lang“ to unnamed module
java·开发语言·dubbo
whyeekkk2 小时前
python打卡第48天
开发语言·python
DougLiang3 小时前
关于easyexcel动态下拉选问题处理
java·开发语言
全职计算机毕业设计4 小时前
基于Java Web的校园失物招领平台设计与实现
java·开发语言·前端
5:004 小时前
云备份项目
linux·开发语言·c++
Eiceblue4 小时前
Python读取PDF:文本、图片与文档属性
数据库·python·pdf
weixin_527550404 小时前
初级程序员入门指南
javascript·python·算法
笨笨马甲5 小时前
Qt Quick模块功能及架构
开发语言·qt