Converting the target molecule into PDBQT

Molecule conversions are a crucial point of AutoDock computations. AutoDock uses PDB format as its base but it cannot make calculations with PDB. AutoDock needs to convert PDB files into PDBQT files. Thankfully it comes with a free AutoDockTools app and if you are lucky, it is easy to make conversions. AutoDock package could drive you crazy while you are trying to make conversions or create configuration files but there are not any other options.

As I said, the app we need to use is AutoDockTools and it is a part of MGLTools  (Molecular Graphics Laboratory Tools) software. Download and install it. Even though I usually work on Mac and Linux I was not able to install MGLTools into any of my UNIX based (Mac or Linux) machines. Mostly the Python version is the problem. MGLTools requires an archaic version of the Python Programming Language. UNIX based systems generally use Python extensively so it is hard to change the system default for MGLTools.

I started my Windows machine and install the Windows version. Installation is pretty straightforward. Just click Next, Next, Next and it is ready to use. The installation wizard will start Python Molecule Viewer at the end but you do not need that. You have to start AutoDockTools via Start menu or the Desktop icon.

After launching the app, you can directly try to open your PDB molecule. File > Read Molecule.

The molecule will be visible on the main window. Edit > Charges > Compute Gasteiger

The resulting warning may be discouraging. But do not worry.

Just click OK and continue with; Grid > Macromolecule > Choose

You will be prompt with the next screen:

Pay attention to the saving address. You have to write the full address of the path including the filename and the extension.

In the end, if you have not experienced any other error you will have a new pdbqt file of your macromolecule.

NEXT: Converting the ligand molecule into PDBQT

Obtaining a Ligand Structure

Sometimes you do not have to create your prospective lead molecule (first ligand molecule to start). In our example, one of the ligand molecules is already in the PDB molecule. It was crystalized while safinamide molecule and we can get it from PDB file.

In the beginning, the strategy of the experiment could be seen a little odd. We are getting a ligand structure form an already docked complex and trying to dock again into the same target! The result should be obvious, right? Yes, that is correct but while we are searching for a new molecule the known lead molecule will help us to find novel therapeutics. Actually, that is why we call it lead molecule.

We will separate the safinamide from MAO-B and re-dock them with a docking tool. While doing that we will be sure about our experiment process. After that make small changes to the lead molecule and invent brand new ligands. After that, the testing (the docking) process will come in handy.

When I review the downloaded target molecule I can see the extra section of atoms. This part is generally indicated with HTATM type. But be careful. FAD is also indicated similarly and it is part of MAO-B. Is it even the binding part of MAO-B. At this point, you have to have some general idea about this target molecule. You can find the reference article which is authored by the scientists who make the crystallography.

Get the safinamide part and save it as ligand.pdb. In the next sections we will concert it into PDBQT format.

NEXT: Converting the target molecule into PDBQT.

Creating the Ligand Structure

In this section of this series, I will describe how to find or make a ligand for molecular docking. When we say ‘ligand’ we mean small molecules which could bind to a larger molecule on a specific binding site. It is mostly easy to draw one and obtain the 3D version of the molecule. Sometimes you may want to extract a ligand structure to test your docking suit.

Drawing the molecule:

I used to use Discovery Studio of Biovia (formerly Accelrys) It has a free visualizer. Visualizer could also be used to draw small molecules. After drawing it you can save the file as PDB, MOL2 or their original file format DSV. Especially I very much like the DSV format but AutoDock Suite is not compatible with it. The only option is PDB. If you use Discovery Studio, you have to save your molecule files as PDB.

Unfortunately, this great software does not have a Mac version and I started to sty other options. Even though there are many other tools I am mainly using a web tool. I do not require any installation of very easy to play with. It is MolView.

You can draw any molecule on the canvas and download the 3D version of your molecule with Tools > MOL file path. The resulting file will not be a PDB file. To be able to convert I prefer Open Babel. Open Babel can be used with the command line and for Mac users, iBabel is the available graphical user interface. Even though I prefer the command line version, you can see the iBabel command screen below. You may skip the Open Babel command sand apply what you see on the screenshot.

Open Babel pretends like it can create PDBQT file for us. Do not believe it. The output will be standard PDB file with PDBQT extension. We will create the PDBQT files both for target and ligand in the next section.

NEXT: Obtaining a Ligand Structure

Obtain The Target Protein Structure

Obtaining the target molecule is crucial. It only depends on the problem you are planning to solve. There are a couple of different possibilities that you can get your protein structure. The easiest one is RCSB Protein Data Bank. If you are looking for a protein molecule which is already scanned and published here, you are free to grab it and use on your experiments.

As you can see on the search box, it is possible to search via molecule name, PDB ID or even with author name. I will use a molecule which I used previously for a research. I will give more details about cleaning and preparing the molecule. But keep in mind that RCSB is not the only source for your target molecules. You can obtain your molecule from somebody who does X-Ray Diffraction or Nuclear  Magnetic Resonance (NMR) for your preferred molecules.

In this case, I am choosing the PDB record of one of the MAO B molecule. Its ID is 2V5Z. After searching with this four character you can visit its page and download PDB file.

After downloading I have the file with name 2v5z.pdb file. Even though it includes everything we need it is not ready to use. First, let’s visualize it with PyMOL. You can download it for any platform.

Published PDB files may include some kind of missing information. But it is possible to add and remove incorrect parts even though we can just do computational operations. For example wavelength of the X-Ray is not suitable to detect all Hydrogen (H) molecules but it is quite easy to calculate correct positions of Hydrogens. Also when you open the molecule with a visualizer, you will see a huge molecule with two subunits and some solvent structure which is the artifacts of the creation process. Let’s check cleanup steps one by one.

Cleanup Steps:

  1. Open your molecule file with PyMOL. You will see the molecule file and the command prompt will be available for commands we have to issue.
  2. Rename the saved file and get rid of redundant information fields of PDB file:
    save molecule.pdb, all

After saving the molecule you can find the file on your root directory of the user. Which could be different on different operating systems. You can use the search functionality of your operating system. molecule.pdb will be the filename if you have not got any error message. Please pay attention to the comma just after the filename.

3. Remove hydrogens by typing exactly this command: (our example may not need this but you will need it for almost all times.)

remove hydrogens

4. Remove solvent by typing exactly this command:

remove solvent

After removing solvent the molecule view will be more clear and small red dots disappear.

4. Select the redundant subunit by typing exactly this command:

sele molecule//B

5. Rename the selection as it is shown in the screenshot. Click A on the “sele” item of the side panel after that you will see “rename selection” on the list. Click it and pay attention to the top left of the molecule view. Just rename it as molecule_subA. Now it is easily selectable. You can do the same for subunit B.

6. Remove oene of the subunit. I will remove B. So I issue this command:

remove molecule_subB
save molecule_subA.pdb, all

(This could  also be done via the side panel)

7. Remove ligand from the molecule.
We are very lucky to have a Safinamide molecule is buried into our file. We will not eliminate it and have the target molecule cleaned up because of the initial structure of the molecule our cavity in the most suitable shape for the docking. We could have saved the file into another pub file but that will be another step of this series so we will just remove it now.

If you open the PDB file with a text editor, you can notice SAG and FAD residues. The FAD is part of the MAO molecules. But SAG is not. It is naturally docked now. Be would like to remove it. Just deleting the SAG lines from the file is enough but I will show the PyMOL way:

sele /molecule_subA//A/SAG
remove sele

8. Save the file. I will rename it as target.pdb since we will use it as out docking target.

save target.pdb, all

In the next section of this series, I will describe who to get the safinamide out of this molecule and other ways to create a ligand from scratch.

NEXT: Obtain or Create the Ligand Structure


A Drug Discovery Walk Through

At the beginning of the 2000s, computational techniques became one of the biggest miracles to pursue a drug discovery journey in an expedited fashion. The technique was not new but arguably the brightest era started around these days. 

The time passed and many full-fledged software packages, even suits of apps, emerged for a faster, more automated and easier solution. Just, I was not convinced about the idea of apps getting easier. That part was always a question on my mind. I was basically thinking of others who are not computer enthusiasts and have no reason to be. Someone might be a great chemist and just wants to work in her own duties. Why should she need to understand how to create a shell script that could run through all molecule files, in a parallel fashion? Even worse; she may have access to a computer with 128 cores waiting idle, but have no idea why the damn simple docking simulation is taking hours.

The day I started my graduate education for Computational Biology and Bioinformatics  I knew that somebody had to fix this mess. I am not sure it is getting easier to use any particular tool to achieve a simple goal. But there are very capable software systems to reach the main goal. Great tools are emerging and software starts to process most of the automation for scientists.  For today, I would like to discuss the current open source tools, then I will describe advances in AI research that could eliminate most of the process clearly.

The main objective is here to show how to do computational small molecule discovery with structure-based drug discovery tools. We will start with obtaining the protein structure data, cleaning and preparing it, how to select a scaffold candidate, prepare multiple variants of a ligand, molecular docking for elimination and at the end analyzing the results.

NEXT: Obtain the Target Protein Structure