Docking - Science topic

Some factors to consider when comparing results from different methods:

1. Methodological differences: Docking and free energy calculations using ONIOM are based on different underlying principles and assumptions. Docking typically predicts the binding pose and affinity of a ligand to a receptor based on scoring functions and geometric complementarity. ONIOM calculations, on the other hand, use a combination of quantum mechanical (QM) and molecular mechanics (MM) methods to describe the electronic structure and energetics of a system. These differences in approach can lead to variations in the absolute values obtained.

2. Relative comparisons: As you correctly mentioned, the primary utility of comparing results from different methods lies in their relative comparisons. The goal is to assess the relative binding affinities or trends between different ligands or receptor-ligand complexes within the same method. For example, if docking experiments consistently rank a set of ligands in a certain order of affinity, and the ONIOM calculations show a similar trend, it provides confidence in the relative comparison between ligands using both methods.

3. Calibration and validation: To establish a meaningful correlation between different methods, it is often necessary to calibrate and validate the results against experimental data or reference datasets. This can involve benchmarking a set of compounds with known binding affinities and comparing the relative rankings obtained from different methods. Calibration can help adjust the scaling factors or parameters to improve the agreement between methods and experimental results.

4. Limitations and assumptions: It's important to consider the limitations and assumptions of each method. Docking, for instance, may not fully capture the precise energetics of the binding process, while ONIOM calculations may have limitations in terms of the chosen level of theory, basis set, or representation of the system. Understanding these limitations can guide the interpretation and comparison of results.

Hope it helps:credit AI

I'm also having the same problem with my predicted model.

Here are a few popular docking tools that allow for multiple-ligand docking:

When dealing with protein-protein docking involving non-standard amino acids, it's important to choose a computational tool that supports these non-standard residues. One such tool that can handle non-standard amino acids is HADDOCK (High Ambiguity-Driven Biomolecular DOCKing).

HADDOCK allows for flexibility in defining molecular interactions, which can be crucial when dealing with non-standard amino acids or modified residues. It supports the use of user-defined restraints, making it adaptable to various types of molecules and molecular features.

Here are the general steps you might follow with HADDOCK:

Prepare Input Files: Prepare PDB files for both interacting proteins, including the non-standard amino acids. Define any additional parameters required for the non-standard amino acids. Define Active and Passive Residues: Specify the residues that are considered active or passive in the binding site. Run HADDOCK: Submit the job to the HADDOCK server or run it locally. Analyze Results: Analyze the docking results, considering the interactions involving non-standard amino acids.

When this type of error occurs to me, I prepare the ligand file again using the openbabel software, without making any modifications... just rewriting it.

example:

obabel -ipdbqt lig.pdbqt -opdbqt -Olig.pdbqt

Typically, such a solution works.

To analyze and validate docking procedure for a protein with no co-crystallized ligand, you need to find the key residue in the active site or binding site of the protein. After choosing the best pose based on docking score and interactions with key residue, you can perform molecular dynamics to validate the docking procedure.

The reasons for control docking/ redocking more than 2 Å might due to:

1. Not included all the important amino acids in the box.

2. The protein protonation at the pocket or ligand charge, but if you follow the standard protocol, it should be no problem.

3. You might need to check for ligand interactions. If there is a covalent bond, you might need to use the covalent bond protocol.

To get the RMSD below 2, you might need to reevaluate the box size/parameters.

What do you mean by the improvment in this term?

Vikas Kumar bro how did you solve this problem? could you please give me your email in the inbox so that I can communicate with you

can anybody please share any research paper that proves that the more the negative score, the better the binding?

@

There can be 2-3 reasons for this. The main reason might be the gridbox not matching with the configuration file.

1. The parameters(center_x, center_y, center_z) in the config file not matching the coordinates set in the grid box - For this you can open the grid.txt file you had prepared and copy the last lines as dimensions for x, y and z in the config file . This will ensure that both the coordinates are matching.

2. Size of parameters in the config file might be incorrect from what is specified by you in the grid - We can set it correctly by opening the grid file and then setting the size in the config file.

Let me know if that works !

Pls don't hesitate to ping me if the problem persists :)

Thank you Waseem Ahmad Ansari and Soykan Agar for your answers. But I could not solve the problem. Can you kindly suggest how to choose the ligand as a static molecule?

Hello Mr. Thomas

Thanks for your valuable comment.

Hello

Here are some common issues and suggestions to troubleshoot:

Hi Kwasi Kantanka Safo ,

This question has been answered before. You can check https://www.researchgate.net/post/How-can-I-parametrize-an-atom-for-Autodock-Tools

If you mean this ClusPro https://cluspro.org/ that you want to use?

It is said clearly that it is a protein-protein docking server. Metformin is not part of protein structure, it is a small molecule ligand.

Arkajyoti Banerjee

My version of YASARA (YASARA-Structure) requires a paid license. However, the price is very reasonable compared to other comparable software such as Schrodinger or CCG MOE. In addition, the licensing policies are extremely liberal. For example, I can run as many simultaneous instances as I wish, on the same computer and/or on multiple computers with the same or different operating systems (it runs on Linux, Mac, Windows, and Android).

There is also a free "viewer" version of YASARA. I have not tried the free version, and I do not know its capabiliites, but I suspect that it would not include docking.

I am glad to know that you got good results with classical AutoDock 4. Sometimes, it works out that AutoDock 4 gives better results than Vina. It depends in part on the particular receptor-ligand system. Standard AutoDock 4 is slower than Vina, although now there is an accelerated version, AutoDock-GPU, than runs on OpenCL or CUDA to achieve very fast speeds.

Hi Fatemeh

maybe you can try HDOCK or PATCHDOCK webservers.

Here are some factors that can contribute to differences in docking results:

Software and Versions: Different docking software packages may use distinct algorithms and scoring functions. Even updates to the same software can lead to differences in results.

Scoring Functions: The scoring functions used to evaluate the binding affinity of ligands to the protein can vary between software and versions. Different scoring functions may prioritize different types of interactions (e.g., hydrogen bonds, hydrophobic interactions) or use different energy terms.

Ligand and Protein Preparation: The way ligands and proteins are prepared for docking, including the addition of hydrogen atoms, assignment of charges, and optimization of structures, can affect the results.

Grid Settings: The definition of the docking grid or search space can have an impact. Slight variations in grid dimensions and placement can yield different results.

Conformational Flexibility: The treatment of ligand and protein flexibility can also influence results. Some docking programs allow flexible ligands or proteins, while others keep them rigid.

Optimization: As you mentioned, optimization of ligands can significantly affect results. Different optimization methods and parameters can lead to variations.

Parameter Settings: Docking programs have numerous parameters and settings that can be adjusted. Small changes in these parameters can lead to differences in results.

Initial Position: The initial position of the ligand within the binding site can also influence the outcome. Docking software often performs multiple runs with different initial positions.

Here's a general process to do this:

Run SwissDock: Start by running your docking simulations in SwissDock to predict the binding of ligands to a protein target.

Export the Results: After the SwissDock simulations are completed, you can typically export the results in a format that is compatible with molecular visualization software like Discovery Studio Visualizer. SwissDock may provide options to export the results in common file formats, such as PDB (Protein Data Bank) or MOL2 (Molecular 3D Structure) files.

Open in Discovery Studio Visualizer: Launch Discovery Studio Visualizer on your computer. Use the "Open" or "Import" feature in Discovery Studio Visualizer to load the exported SwissDock results. This usually involves selecting the PDB or MOL2 files containing the protein-ligand complex structures.

Analyze Ligand Interactions: Once you have imported the SwissDock results into Discovery Studio Visualizer, you can analyze the ligand interactions with the protein target. You can visualize the binding modes, hydrogen bonds, hydrophobic interactions, and other molecular interactions between the ligands and the protein.

Visualize and Interpret: Use the visualization and analysis tools in Discovery Studio Visualizer to examine the docking results, study the ligand-protein interactions, and gain insights into the binding affinity and stability of the complexes.

Hello Smruti,

I had the same problem and I solved it by updating the "Configuration" TAB

After I did that, it seems all the plugins need it for each docking method uploaded into "DockingPie".

Please try that

Hi Mohd Sufiyan Ansari ,

I noticed that you named your folder "Project Docking" which contained a space between the word.

Therefore, rom your screenshot, it stated that it can't find or open Grid Parameter File "C:/Project" means it reads your folder as Project instead of Project Docking.

My suggestion is to remove the space in your folder name (ProjectDocking). restart ADT, and make sure to set your folder as the startup directory (this is essential whenever using AutoDock or AutoDockTool in Windows).

Hope this help. All the best.

In a simple answer, It's judged on a case-by-case basis. Hetatoms (non-protein atoms) that make important contacts with potential ligands or impact the structure/dynamics of the binding site should be kept. Less critical ones can be removed to simplify the system.

Here are a few key points to consider when deciding which hetatoms to keep when preparing a protein receptor for molecular docking:

- Metal ions that are important for structure/function should generally be kept. For example, zinc ions that are important for holding enzyme active sites in the proper conformation.

- Common cofactors like heme groups, flavins, and nicotinamide should also be kept if they are present in the binding site. They are often directly involved in ligand interactions.

- Structural ions like calcium and magnesium ions that help stabilize protein structure are usually okay to remove unless they make direct contacts with the ligand. They tend to make the receptor more rigid.

- Bulk solvent molecules are typically removed to make the binding site more accessible to ligands.

- Other heteroatoms like glycosylation groups may be kept if they reside in the binding site and could impact ligand binding. Those distant from the site can be removed.

- Prosthetic groups or other modulators that define the binding site should be retained. Those not involved can be removed.

Thank u. I will try doing this

Dear Michael

Thank you for the suggestion. We successfully resolved the problem

The selection of a protein for molecular dynamics (MD) simulations based on AutoDocking results should consider several factors:

Glide and MM-GBSA (Molecular Mechanics/Generalized Born Surface Area) are widely employed techniques in molecular docking and the computation of binding free energies. To acquire Glide docking scores and MM-GBSA scores, you can follow these general procedures:

It's crucial to remember that the specific steps and commands may vary depending on the software being employed. The Schrödinger suite, including Glide and Prime, is commonly utilized for these calculations. Additionally, the interpretation of Glide docking scores and MM-GBSA results should be approached with caution, as they provide estimates and may not always precisely predict binding affinities. Validation and comparison with experimental data are essential for assessing the reliability of these calculations.

Amanpreet Kaur Thank you for your information. Could you provide a reference for free-binding protein (-8 kcal/mol), please? I will drive into it.

All the best,

Ramkumar Katturajan Hicham Mechqoq sir will you please tell me how to do merging of files

With individual dock parameter files of the inhibitor molecules, substrate’s dpf was merged into one single file to run MLSD simulation??

Sneha Sneha thank you so much. I will try it.

This could be due to that the program uses different starting positions for the ligands in the simultaneous docking case, which could lead to different conformations and interactions with the receptor. It is also possible that the ligands are competing with each other for the same binding site, which could affect their binding energy and interacting amino acids.

Good luck.

Sethupathi Raj S Did you get any procedures?

Thank Ayaz Anwar & Pawan Kumar for your inputs

Jamoliddin Razzokov Sir, i put the ligand-protein dock pdb format in solution builder of charmm gui and tick the heteatom and next is search csml/mol2 to generate pdb. After this the charmm terminated abnormally. Should i separate protein and ligand from protein-ligand docked pdb?

Thank you!

You can add this line in the AD4 parameters.dat file

Regards

MOE is also a very reliable software for protein-ligand docking, I've used it several times and I use Schrodinger too. So u can use it if you need an alternative to Schrodinger

there was probably an error in defining the protein and ligand molecules. You can try to check the structure manually on Notepad for errors

The term "good" is relative. Just like Dr. Rajesh Pal mentioned, one strategy is to dock another compound which is either a known binder (ligand) or a known non-binder. Then you can compare if -5 Kcal/mol is "good" binding energy or not.

As a general rule, ligands that binding successfully to deep binding pockets or cavities have large-negative binding energies. While ligands that bind successfully to shallow pockets (for example during protein-peptide interactions) tend to have small-negative binding energies.

Hi,

the best way is to do it already in the database. Create a database with all you ligands and then Compute -> Molecule -> Wash. Here you define protonation states and all the rest.

Hi,

you have to add this to either the AD4.1_bound.dat or AD4_parameters.dat (often found inside MGLTools directory):

atom_par Pd 1.34 0.048 12.000 -0.00110 0.0 0.0 0 -1 -1 4 # Non H-bonding

Also, as it is VINA, you might have to recompile it again.

Hello Piyush. I am not able to completely understand your problem. Did you download a protein with an ion "nd" that you want to re-dock with using pyrx? Or did you separately downloaded the ion file and want to perform docking with the unbound protein?

Hello Samrat Paudel Samrat Paudel, I am facing the same issue and I can't download the subsets I need for my docking study. If it's okay with you, I would greatly appreciate it if you could share the compounds you've downloaded with me. Best regards.

Chandni Pathak

Save your docking complex as a PDB file. Open the file in a text editor (not a word processor) such as gedit for linux or Notepad++ for Windows. Scroll down near the end of the file to the "HETATM" section and look at the atom names for the atoms in the ligand. Do the same with the PDB file of the X-ray structure of the complex. The names of the corresponding atoms in the ligand in the docking complex and the ligand in the X-ray structure need to be the same for doing an RMSD calculation. To help with recognizing the atoms in the ligand, you could open the docking ligand and the X-ray ligand in a molecular viewer. You can manually edit the ligand atoms in one of your PDB files using a text editor so that the corresponding ligand atoms have the same atom names. Then save the file as a PDB file. If all goes well, you can then align the two PDB structures in a molecular viewer and carry out the RMSD calculation. I hope this helps. Good luck!

-- RJR

Dear Yahaya,

Thank you for your insightful suggestion regarding the active site prediction tool from the Supercomputing Facility for Bioinformatics & Computational Biology at IIT Delhi. I appreciate your thoroughness in exploring the potential utility of this tool.

I'm glad to make your attention to the paper by Tanya Singh, D. Biswas, and B. Jayaram titled "AADS - An automated active site identification, docking and scoring protocol for protein targets based on physico-chemical descriptors." The fact that this paper has been cited by 181 researchers highlights its significance in the field. While I haven't personally used the server yet. Though a humble suggestion to leverage it for gaining a preliminary understanding of the active site of proteins of interest is intriguing.

I would love to suggest an approach of cross-checking the tool's predictions with known active sites of a minimum of 10 proteins from the same organism is a thoughtful idea. This comparative validation would indeed provide valuable insights into the reliability and accuracy of the tool's predictions. If the results align well with established knowledge, it would lend substantial credibility to the tool's effectiveness.

Considering the extensive citations of the paper and the potential of the tool, I believe that exploring its feasibility is a worthwhile endeavor. The ability to gain a preliminary idea about the active site of proteins of interest could greatly inform and streamline any research efforts.

Additionally, there are several other active site prediction tools available that you might find useful for your research. Here are a few notable options:

Hope this answer will be helpful for you.

Hi Julián Santiago Ruiz Castellanos,

In addition to Waseem Ahmad Ansari, molecular docking provides the initial prediction of the binding poses of a ligand against the target protein, also giving information on putative binding sites in the protein and significant residues involved in the ligand mode of action.

After this, one can select the best binding pose as input for molecular dynamics simulations for effective results.

Docking and simulations are complementary techniques, together providing a better understanding of the ligand-protein interactions.

Abisha Meji Milon I don't think you can use gaussian for docking studies.

Try autodock vina.

A recently just-released publication pointed out that while AutoDock Vina is faster, AutoDock 4 correlates better with experimental binding affinity.

To quantify the number of residues and minimum distance from a multi-docking result in a single Python script, you can use a molecular docking package such as AutoDock Vina or PyRx, and a molecular dynamics package such as MDAnalysis.

Here's a general approach to accomplish this:

1. Load the multi-docking result into a molecular docking package such as AutoDock Vina or PyRx. These packages can read in the protein and ligand structures and perform the docking calculations to generate multiple docking poses.

2. Use MDAnalysis to analyze the docking poses and calculate the number of residues and minimum distance for each pose. MDAnalysis is a Python library that can be used to analyze molecular dynamics trajectories, including molecular docking results.

3. Write a Python script that uses MDAnalysis to loop through each docking pose and calculate the number of residues and minimum distance for each pose. Here's an example code snippet that uses MDAnalysis to calculate the minimum distance between the protein and ligand in a single docking pose:

```python

import MDAnalysis as mda

# Load the protein and ligand structures

protein = mda.Universe("protein.pdb")

ligand = mda.Universe("ligand.pdb")

# Load the docking pose

pose = mda.Merge(protein, ligand)

pose.load("docking_pose.pdb")

# Calculate the minimum distance between the protein and ligand

protein_atoms = protein.select_atoms("protein")

ligand_atoms = ligand.select_atoms("not protein")

min_distance = mda.analysis.distances.distance_array(protein_atoms.positions, ligand_atoms.positions).min()

# Print the number of residues and minimum distance for the docking pose

num_residues = len(protein_atoms.residues)

print(f"Number of residues: {num_residues}")

print(f"Minimum distance: {min_distance}")

```

4. Modify the script to loop through each docking pose and calculate the number of residues and minimum distance for each pose. You can use a for loop to iterate through each docking pose and store the results in a list or array:

```python

results = []

# Loop through each docking pose

for i in range(num_poses):

# Load the docking pose

pose = mda.Merge(protein, ligand)

pose.load(f"docking_pose_{i}.pdb")

# Calculate the minimum distance between the protein and ligand

protein_atoms = protein.select_atoms("protein")

ligand_atoms = ligand.select_atoms("not protein")

min_distance = mda.analysis.distances.distance_array(protein_atoms.positions, ligand_atoms.positions).min()

# Calculate the number of residues

num_residues = len(protein_atoms.residues)

# Store the results in a tuple

results.append((num_residues, min_distance))

# Print the results

for i, (num_residues, min_distance) in enumerate(results):

print(f"Docking pose {i}: Number of residues = {num_residues}, Minimum distance = {min_distance}")

```

This is just a general approach, and the specific details of the script will depend on the molecular docking and molecular dynamics packages being used, as well as the format of the multi-docking result.

Hope this helps!!

Credit : mainly to AI tools

Add this to the dat file....

atom_par Ir 2.84 0.073 12.000 -0.00110 0.0 0.0 0 -1 -1 1 # Non H-bonding

Regards

Contact Schrondinger and request for free trial version. The whole package is just awesome and covers many computational parameters.

If you have linux system, then you will also be able to perform MD simulation in the same package. Good luck!

I had the same issue which made me exhausted while doing flexibke docking. Thank God, got the free trial version of schrodinger package which solved many problems. I will suggest you to plz change your approach

the ligand bind but not inside the grid box. I think I find out why it happened.

If you just want to group the files into a single .pdbqt just

cat *.pdbqt > group.pdbqt

You can use autodock vina. Separate the ligand and protein from each other and prepare them for docking. While running the vina, use --score_only option which only scores the current crystal pose and does not try to predict new pose.

I can think of several reasons why the docking result may not translate to an in vitro or in vivo effect.

1. The docking results do not adequately reflect reality. Docking scoring functions are notoriously inaccurate.

2. The compounds bind to the target, but do not cause any inhibition because they do not interfere with the biological function. This could happen if the binding site of the compounds is outside the active site or site of protein-protein interaction.

3. The compounds are too insoluble to reach the target in an aqueous environment. This is likely to be the case for long-chain hydrocarbons.

4. The compounds are too toxic to be used at an effective concentration.

5. The compounds are too metabolically labile and/or have too little bioavailability to reach an effective concentration at the site of the target.

Did you minimize the protein???

If not, please minimize the protein prior to using for any analysis..

From the reviewer's question, it seems they are interested in gaining a deeper understanding of the different interaction mechanisms involved in the docking process and how they contribute to the overall score function. The score function is a key component in docking software that evaluates and ranks different ligand-protein interactions based on their predicted binding affinity.

It's unclear what you mean by redocking and validation.

HADDOCK is not for docking any metal ions to proteins. Neither any other docking tools supports that.

You can model the Zinc ion coordination complex to a protein (say using modeller) and run MD simulation for a minimum of 100ns to validate and restrain/refine the model. Also, Sometimes you might have to model the surrounding residues/loops so that it can form a proper coordination complex.

Hello @Mazedul can You remember how you went about this problem??

Hello Rubem, thank you for answering my question. However i tried the method that you pointed out and failed to make work, error message "NameError: name 'rms_cur' is not defined" pops out. I apologise for the inconvinience but can you help me with help me further?

The relationship between ligand size and binding affinity in silico docking can be complex and depends on several factors. In general, larger ligands are more likely to interact favorably with the binding site due to the greater contact area and flexibility. However, the size of the binder can also lead to steric challenges and difficulty in fitting. Binding affinity depends not only on size, but also on the chemical composition and properties of the ligand. Each system must be analyzed individually, considering other factors, such as physical-chemical properties and characteristics of the binding site.

Mahsa Alem The error message you encountered during the macromolecule preparation indicates that some atoms in your input file do not have the required "autodock_element" field. This field is necessary for creating the PDBQT file used in docking simulations with the Python molecular viewer.

To resolve this issue, you will need to ensure that all atoms in your macromolecule have the "autodock_element" field properly defined. Here are a few steps you can take to fix the error:

1. Check your input file: Review the file you provided as input for macromolecule preparation. Make sure that each atom entry includes the "autodock_element" field and that it is populated correctly. This field should correspond to the elemental symbol of the atom (e.g., C for carbon, O for oxygen).

2. Update the input file: If you find any missing or incorrect "autodock_element" fields, you need to update the input file accordingly. Open the file in a text editor or a molecular visualization software capable of editing PDB or PDBQT files. Locate the problematic atoms and add or correct their "autodock_element" values.

3. Save and reprocess: After making the necessary changes to the input file, save it with the modifications. Then, rerun the macromolecule preparation process using the Python molecular viewer. The updated file should now contain the required "autodock_element" fields for all atoms, allowing the creation of the PDBQT file without encountering the previous error.

It's important to carefully review the documentation or user guide provided with the Python molecular viewer to ensure you are following the correct syntax and formatting requirements for defining the "autodock_element" field. If you're unsure about the specific steps to update the input file or encounter any further issues, you may also consider reaching out to the software developers or community support for assistance.

For batch docking, I have used the vLifeMDS tool.

Hi Sunidhi,

I am not a Gromacs user, but the error seems to indicate that the atomtype placement is incorrect. You can refer to the following tutorial for more information:

Regards,

Aashish