fragment_processor.py

import os
import pandas as pd
from pathlib import Path
from mmpdblib.fragment_io import read_fragment_records
from rdkit import Chem

def fragmentize_molecule(smiles_string, max_ratio=0.5):
    # 创建临时文件名
    input_file = "temp_input.smi"
    output_file = "temp_output.fragments"
    
    # 将SMILES字符串写入临时输入文件
    with open(input_file, "w") as f:
        f.write(smiles_string + "\t" + "Molecule" + "\n")
    
    # 使用mmpdb工具进行分子碎片化
    os.system(f"mmpdb fragment {input_file} -o {output_file}")
    
    # 读取并处理碎片
    fragment_reader = read_fragment_records(output_file)
    fragment_list = []
    
    for record in fragment_reader:
        for frag in record.fragments:
            if count_heavy_atoms(frag.variable_smiles) < count_heavy_atoms(record.normalized_smiles) * max_ratio:
                fragment_list.append({
                    'variable_smiles': frag.variable_smiles,
                    'constant_smiles': frag.constant_smiles,
                    'record_id': record.id,
                    'normalized_smiles': record.normalized_smiles,
                    'attachment_order': frag.attachment_order
                })
    
    # 删除临时文件
    os.remove(input_file)
    os.remove(output_file)
    
    # 返回碎片列表
    return pd.DataFrame(fragment_list)

def count_heavy_atoms(smiles):
    # 使用RDKit计算重原子数
    mol = Chem.MolFromSmiles(smiles)
    return mol.GetNumHeavyAtoms() if mol else 0

# 示例调用
# smiles = "O=C1CCCC2=C1C1(CCS(=O)(=O)C1)N=C(Nc1nc3ccccc3o1)N2"
# fragment_df = fragmentize_molecule(smiles)

# print(fragment_df)