# Import required packages import os from Bio.PDB import PDBParser, PDBList from numpy import * import pandas as pd from pymol import cmd from ImmuneBuilder import ABodyBuilder2 # Amino acids three letter to one letter dictionary amino_acids = { "ALA": "A", "ARG": "R", "ASN": "N", "ASP": "D", "CYS": "C", "GLU": "E", "GLN": "Q", "GLY": "G", "HIS": "H", "ILE": "I", "LEU": "L", "LYS": "K", "MET": "M", "PHE": "F", "PRO": "P", "SER": "S", "THR": "T", "TRP": "W", "TYR": "Y", "VAL": "V" } def get_pdb_file(identifier, overwrite_files=False): """ Download PDB file from the protein data bank and save it in structures folder as .pdb Args: identifier (Str): 4-letter PDB identifier overwrite_files (Bool): Overwrite already downloaded structure, default=False Returns: None """ # Check if structures folder exist if not os.path.isdir("./structures/"): os.mkdir("./structures/") # Check if file already exist if os.path.isfile("./structures/" + identifier + ".pdb") and not overwrite_files: return None # Else-if PDB file already exists, remove this file elif os.path.isfile("./structures/" + identifier + ".pdb") and overwrite_files: os.remove("./structures/" + identifier + ".pdb") # Instantiate PDBList pdbl = PDBList() # Ask BioPDB.PDBList to get PDB file pdbl.retrieve_pdb_file(identifier, file_format="pdb", pdir="./structures/", overwrite=overwrite_files) # Rename the downloaded file os.rename("./structures/pdb" + identifier + ".ent", "./structures/" + identifier + ".pdb") # Return None return None def get_chain_from_pdb(pdb_file, chain_names, choosen_model=0): """ Extract a specific chain's sequence from a PDB file Args: pdb_file (Str): Name and location of PDB file chain_names (List): List chain(s) name(s) to extract choosen_model (Int): Which model number to get chain(s) from Returns: Dictionary of sequences for chains """ # Get the PDB id from filename pdb_id = os.path.basename(pdb_file) # Instanciate PDBParser parser = PDBParser(QUIET=True) # Get structure and specified model of protein from PDB file structure = parser.get_structure(pdb_id, filename) model = structure[choosen_model] # Dictionary to store found chain chain_sequences = {} # Loop through the list of specified chain names for chain_name in chain_names: # List of found residues in current chain residues = [] # Find all residue names in chain try: for residue in model[chain_name]: # Get the name of the current residue res_name = residue.get_resname() # If the residue is an amino acid, we add it to the list of residues for the chain if res_name in amino_acids.keys(): residues.append(amino_acids[res_name]) # If the chain doesn't exist we suppress the error and continue except KeyError: break # Put the found chain sequence into the list of found sequences chain_sequences[chain_name] = "".join(residues) # Return the chain sequences found return chain_sequences if __name__ == "__main__": # Data set antibodies = ["8ahn", "8cwi", "8f6l", "8bse", "7wsl", "8ffe", "7ox3", "8dfi", "7str", "7vke", "4n1c", "4xvt", "7vux", "7amr", "6mvl", "6svl", "6wzk", "5bjz", "5ggv", "3sob", "3nps", "6k64", "6kyz", "5n88", "5hi4", "6sge", "7kql", "8dce", "7mzg", "5ggs", "6k3m", "5u5m", "7jum", "7nfr", "3p0y", "7jmp", "7mdp", "6cbv", "6k6a", "7z0x", "6ba5", "5wk3", "4fqi", "4ydl", "3mxw", "7e9b", "6ban", "5vkd", "5uea", "3l5x", "5u3d", "2vxq", "7so5", "6b0g", "7lfb", "4xmp", "7lm8", "4tsb", "7bz5", "5ob5", "6vy4", "6aod", "6was", "5umn", "7kmh", "7q0g", "7neh", "5uek", "4h8w", "3d85", "6nmt", "6k0y", "6phb", "4hjg", "5u5f", "6a3w", "5ucb", "4m62", "7b3o", "6dkj", "7mzi", "7n4n", "6qb3", "4nzr", "7rxp", "6qfc", "7lm9", "4ioi", "2nxy", "5n7w", "7sd5", "6b0s", "6iea", "5l6y", "7eow", "5ngv", "2yc1", "3u2s", "2uzi", "2xwt", "2ny2", "5w0d", "6e56", "6o39", "5u6a", "5vag", "4dtg", "4i77", "7kmi", "7n3c", "3se9", "7fcq", "7lsg", "6meh", "7d6y", "3se8", "5m2j", "4al8", "4j6r", "7d5b", "4xvs", "7n3d", "7rp3", "5sy8"] # Dictionary of accepted PDBs from list above accepted_pdbs = {} # Loop through all antibody PDBs specified above for antibody in antibodies: # Download PDB files using BioPDB get_pdb_file(antibody) # Construct file name variable of save location filename = "./structures/" + antibody + ".pdb" sequences = get_chain_from_pdb(filename, ["H", "L"]) # If we were able to get a H and L sequences from PDB, the we accept it # Unfortunately, some of the PDBs in the dataset have named light and heavy chains other names than H and L # I omit these from the exercise if "H" and "L" in sequences.keys(): accepted_pdbs[antibody] = sequences # Create an output directory, if it does not exist output_location = "./predictions/" if not os.path.isdir(output_location): os.mkdir(output_location) # Predicted structures predicted_structures = [] # Loop through all accept PDB files of antibodies for antibody in accepted_pdbs.keys(): # Set path of output file output_file = "./predictions/" + antibody + ".pdb" # If the structure have already been predicted, skip if os.path.isfile(output_file): predicted_structures.append(antibody) continue # Instanciate ABodyBuilder2 antibody_builder = ABodyBuilder2() # Predict antibody and save it to the output file try: predict_antibody = antibody_builder.predict(accepted_pdbs[antibody]) predict_antibody.save(output_file) predicted_structures.append(antibody) # In case we get an error from the prediction, we move on to the next antibody except AssertionError: continue # Dictionary containing RMSD values of superpositioning predicted structure onto crystal structure rmsd_values = {} # Calculate RMSD of superpositioning predicted structure on crystal structure # Loop through all predicted structures for structure in predicted_structures: # Set object name of structures for use in PyMol crystal_name = "crystal_" + structure prediction_name = "prediction_" + structure # Set crystal and prediction file paths crystal_pdb = "./structures/" + structure + ".pdb" prediction_pdb = "./predictions/" + structure + ".pdb" # Load structures into PyMol cmd.load(filename=crystal_pdb, object=crystal_name) cmd.load(filename=prediction_pdb, object=prediction_name) # Select H and L chains select_crystal = crystal_name + " and chain H+L" select_crystal_name = crystal_name + "_chains" select_prediction = prediction_name + " and chain H+L" select_prediction_name = prediction_name + "_chains" cmd.select(select_crystal_name, select_crystal) cmd.select(select_prediction_name, select_prediction) # Superposition predicted structure onto crystal structure superposition = cmd.align(select_crystal_name, select_prediction_name, object="Superposition") # Save RMSD value to dictionary rmsd = superposition[0] rmsd_values[structure] = rmsd # Delete everything in PyMol before we continue cmd.delete("*") # Save RMSD values to file using Pandas DataFrame rmsd_df = pd.DataFrame.from_dict(rmsd_values, orient="index", columns=["RMSD"]) rmsd_df.to_csv("./protein.csv", sep=";", index_label="PDB_ID")