💻Easy way to classify PDB ligand-protein complexes depending on covalent or non-covalent ligand

PDB-CAT is a Jupyter Notebook that aims to automatically categorize the PDB structures based on the type of interaction between atoms in the protein and the ligand

Last updated 4 months ago

💻Easy way to classify PDB ligand-protein complexes depending on covalent or non-covalent ligand

PDB-CAT is a Jupyter Notebook that aims to automatically categorize the PDB structures based on the type of interaction between atoms in the protein and the ligand

PDB-CAT is a program that classifies a group of protein structures into three categories: free-ligand, covalent bonded, and non-covalent bonded. Besides the classification, the program can verify if there are any mutations in the protein sequence by comparing it to a reference sequence

Open in Colab

Execute the Environment setup by clicking on the ▶️ button.

👈 Add your files to the cif folder

Mutation filter, True or False?

If using the mutation filter, fill in the variable reference with the file name and set the mutation to True. For example:

reference = "1ABC.cif"
mutation = True

Make sure all cells have been executed by playing the ▶️ button

Download results in a zip

And that's it! If your computer downloads the zip file containing the results, the program has been correctly executed.

🚀 Learn how to use it with this example!

SARS-CoV-2 Main Protease

We will begin downloading CIF files and saving them in a folder named /cif.

Download All Structures

You can use this method:

Download the structures of known IDs with the following executable:

https://www.rcsb.org/docs/programmatic-access/batch-downloads-with-shell-script

-f specify input file with id separate by comma

-c for cif-gz file

-o specifiy output path

 ./batch_download.sh -f input.txt -c -o cif # execute to download by ID names
  gunzip *.gz # Decompress downloaded .gz

Or:

Download cif directory.

If you are familiar with GitHub you can clone the following repository:

  git clone https://github.com/URV-cheminformatics/PDB-CAT.git

✍️How to Start in 3 SIMPLE STEPS

It is already Downloaded, but you can take a look here!

The blacklist compiles more than 280 codes for solvents, ions, co-factors, and other substances capable of bonding with the protein structure. This information is stored in a text file that users can edit, allowing for the inclusion of new codes or adjustments related to the significance of co-factors and solvents in the analysis.

Open the Jupyter Notebook! The first cell looks like this:

from PDBCAT_module import *
from pdbecif.mmcif_io import CifFileReader
from pdbecif.mmcif_tools import MMCIF2Dict
import pandas as pd
import time
import re
import os
import shutil
from Bio.Align import PairwiseAligner 
from Bio.PDB import * 
from datetime import datetime

Download requirements.txt and save it in the same directory. Execute the following code in the terminal:

pip install -r requirements.txt

You only need to modify:

mutation

Choose to use the mutation filter

Mutation = True (useful in case you are using the same protein dataset of structures)

Mutation = False (useful in case you are working with different proteins)

pdb_reference_sequence

Depending on your target, you will need to change the code name.

▶️ Run the code!

🏁Output

📄 CSV

In the first CSV, a line is written for each PDB ID code, providing a comprehensive set of information. This section includes details related to the protein, such as the PDB ID, title of the PDB file, protein description, number of subunits, subunits ID - referred as chain -, and the number of residues for each subunit. Subsequently, it indicates whether it is a complex. Following this, information about discarded ligands - elements in the blacklist bonded to the protein - and branched molecules their names, types, functions, and the presence of a covalent bond is provided. Next, ligand information is presented, including the name, type, functions, and the presence of a covalent bond. The final segment covers mutation information, specifying the number of mutations, their location, identity percentage, and gaps.

In the second CSV, a line is written for each entity bonded to a protein. It is straightforward, containing the ID of the protein, the bonded molecule, its name, type, function, and whether it is covalently bonded and, if so, with which residue. Additionally, if it is a glycosylation, that information is also included.

🗂️ Folder Classification

The program generates a set of folders to separate ligand-free structures, covalent complexes, and non-covalent complexes. Only covalent bonds between ligands and proteins are considered for classifying a complex as covalent. If the mutation filter is used, this classification is performed within the non-mutated folder, and the mutated folder is created alongside the non-mutated one.

Out of the 1,335 Mpro structures downloaded from the PDB, 1,128 were identified as non-mutated. These structures were further classified into 100 ligand-free structures, 399 covalent complexes, and 629 non-covalent complexes. Additionally, among the covalent complexes, 27 were specifically identified as peptide ligands.

❓How does the code classify the files?

Last updated 4 months ago