PiLAH

PiLAH stands for Protein Ligand Abstraction and Hydrogenization tool, it is a cheminformatic tool for protein and ligand preparation. Given a PDB file it will separate the protein and the ligand, fix the bond order of the ligand, protonate protein and ligand separately, then write them to separate file.

It uses Biopython, RDKit, Meeko, and Openbabel for the general cheminformatics processing, then it will use Dimorphite-DL and pKAI to protonate small molecule and protein respectively.

Features

Ligand Processing

• Choose ligand using its chain ID and residue ID.
• Bond order correction using SMILES template.
• Bond order correction for ligand with missing atoms.
• Protonation using Dimorphite-DL.
• Choose the pH environment.

Protein Processing

• Choose protein using its chain ID (multiple chain is possible).
• Able to include/exclude metal.
• Removing residue with missing atoms.
• Removing residue with incorrect bond length or angle.
• Renumbering residues for structure with Insertion Code.
• Choose residue coordinate with Alternate Location (altloc id).
• Protonation using pKAI model.
• Choose the pH environment.

Input Output

• Input in PDB format.
• Ligand output in PDB, MOL2, SDF, and PDBQT format
• Protein output in PDB, MOL2, and PDBQT format.
• 2D image (PNG or SVG format) of generated ligand for easier inspection.

Installation

Download the binary release from the release page then extract it. In the extracted folder you will find an executable file named with pilah and an example folder which contain configuration and input files for the following usage guidelines.

You can execute PiLAH directly from the extraction folder as such:

./pilah --help

It is highly recommended to copy PiLAH to one of your executable path. You can check your available executable path by running the command echo $PATH. Some path that you can use are /home/$USER/.local/bin or usr/local/bin.

After copying PiLAH to your executable path, go to PiLAH's example directory and run PiLAH again:

pilah --help

If it runs properly, the next step is to run PiLAH using one of the provided example. Still in the example directory run PiLAH with the following command:

pilah run config_pdb_hux.txt

Normally it will take between 5-15 seconds for PiLAH to complete the protein ligand extraction and protonation. After PiLAH finished the processing, PiLAH will generate four files. One for the isolated protein molecule, protein_1e66.pdb. Second one for the extracted ligand molecule, HUX.pdb. Third one for the ligand image in 2D representation, HUX.png. The last one for the log file which is named with the date and time when PiLAH was executed, such as log_20240815_172113.txt

Now PiLAH is ready to be used on your system.

Getting Started

To get started we will use the crystal structure of HDAC8 complexed with Quisinostat (PDB ID: 6HSH).

Mandatory Options

The PDB file of this structure is already included in the examples folder. To extract the protein and the ligand from the PDB file we will need a configuration file that must contain the following mandatory options:

1. input : the PDB file name
2. ligand_id : the residue name of the ligand inside PDB file
3. protein_chain : The chain ID of the selected protein
4. ligand_chain : The chain ID of the selected ligand
5. ligand_smiles: The SMILES of the ligand, which will be used as the molecule template for the ligand to make sure every bond order in the ligand correctly assigned.
6. protein_out : the output filename for the protein
7. ligand_out : the output filename for the ligand

First we will use the config_pdb_gok.txt configuration file.

input = 6hsh.pdb
ligand_id = GOK
protein_chain = A
ligand_chain = A
include_metal = yes
protein_out = protein_6hsh.pdb
ligand_out = GOK.pdb
ligand_smiles = Cn1cc(c2c1cccc2)CNCC3CCN(CC3)c4ncc(cn4)C(=O)NO
ph = 7.4

The input, protein_out, and ligand_out are self-explanatory. The protein_chain and ligand_chain options are for selecting the chain ID. The chain ID option for protein and ligand have to be provided separately because there are many cases where the protein and the ligand were located at different chains. The ligand_id option assigned with GOK which is the residue name of Quisinostat as can be seen in the PDB page of HDAC8-Quisinostat complex.

Ligand SMILES for the ligand can be retrieved from the corresponding PDB page, specifically in Small Molecules section. And as can be seen below there is a link for GOK:

If we open the link above we can find the SMILES for Quisinostat in Chemical Component Summary table

As we can see the SMILES in the GOK page above is the same as the SMILES in the configuration file above. Next, lets try extracting the protein and ligand from PDB file using the configuration file above and analyze the result.

Optional Options

Other than the seven mandatory options above, there are optional options:

1. include_metal : include metal as part of the extracted protein structure. Only accept yes and no value, the default is no.
2. ligand_image : the image file name for ligand 2D representation
3. image_size : the image size of the ligand image (explained below)
4. ph : the pH value for determining the protonation state of ligand and protein residues. When not provided, the default value is 7.4.
5. ptreshold: the minimum difference between pH and pKa to allow the ionizable residues to be ionized. The default is 1.0
6. pkai_model: the pKa model that will be used by pKAI, the default is pKAI
7. ligand_res_num: selected ligand residue number when multiple ligand with the same ID exist in one chain.
8. alt_loc: selected alt loc id for disordered residue.

Many of these optional options will be explained in details in Various Use Cases section in the documentation.

Running PiLAH and Analyze the Results

In the examples directory execute pilah run config_pdb_gok.txt, or if you have not copied PiLAH into your executable path as instructed on installation instruction above you can run PiLAH directly from the higher directory using the command ../pilah run config_pdb_gok.txt.

After PiLAH finished running there should be three files generated:

1. protein_6hsh.pdb : The structure of HDAC8 chain A.
2. GOK.pdb : The structure of Quisinostat chain A.
3. log_20240816_210348.txt : The log file which contain the configuration being set up and the ionization records for each ionizable residue of the protein.

Last but not least, the log file contain important information about the PiLAH version, configuration, and ionization records:

Ionization records is a list of ionizable protein residues with its predicted pKa value and charge. According to the author of pKAI there are four possibly negatively charged residues (ASP, GLU, CYS, TYR) and three possibly positively charged residues (LYS, ARG, HIS). However it is important to note that the ARG residue considered to be always positively charged and therefore not calculated by pKAI. So we deliberately decided to include ARG in ionization records and assign 14.0 as the pKa value and the charge is always Positive. And we gave the same treatment with any metal ion.

Therefore, behind the scene pKAI will calculate the pKa for all ionizable residues except ARG. And then as general rule, for ASP, GLU, CYS, and TYR, ionizable residues will be neutral when the pH below the pK, else they will be negatively charged when the pH above the pK. For LYS, ARG, and HIS, ionizable residues will be neutral when the pH above the pK, else they will be positively charged when the pH below the pK. And as general rule, the ionizable residues only get ionized when the pH-pK difference is 1 unit or more. This difference can be set up with ptreshold option as explained in Various Use Cases in the documentation.

Generating 2D representation of the ligand

We can also view the 2D representation by adding the ligand_image and ligand_size option to config_pdb_gok.txt as shown in the following config file:

input = 6hsh.pdb
ligand_id = GOK
protein_chain = A
ligand_chain = A
include_metal = yes
protein_out = protein_6hsh.pdb
ligand_out = GOK.pdb
ligand_image = GOK.png
image_size = large
ligand_smiles = Cn1cc(c2c1cccc2)CNCC3CCN(CC3)c4ncc(cn4)C(=O)NO
ph = 7.4

which generate image file GOK.png:

By inspecting the image we can immediately figure out the protonation state of each ionizable moiety. In the image above we can see that the dimethylamino moiety get protonated at pH 7.4, if you are curious you could re-run this example after modifying the pH value to 9.4 and see the difference.

What is next?

The example that we use above should cover most of your need. For other use cases, such as how to generate other molecule format, how to use different protonation model or environment, and so on please check how to use PiLAH in various use cases in the documentation.

License

PiLAH was created by Muhammad Radifar and Enade Perdana Istyastono. It is licensed under the terms of the Apache License 2.0 license.

Credits

PiLAH was created with cookiecutter and the py-pkgs-cookiecutter template.