A Bilingual Computational Workflow for Identifying Potential PLK1 Inhibitors in American Sign Language and English

A bilingual computational workflow.Introduction. Research shows PLK1 is overexpressed in various cancers. Inhibition of PLK1 through its polo-box domain, PBD, has been shown to induce apoptosis.

The goal of this study is to identify selective PLK1-PBD inhibitors with desired medicinal and pharmacokinetic properties. Homologs PLK2 and PLK3 were subjected to the same protocols to assess selectivity.Protocol. Target protein preparation.

Here we utilize CHARMM-GUI to optimize PLK1, 4HCO, from the Protein Data Bank, PDB. This ensures that the protein does not have any missing residues or hydrogens. Let's proceed with the tool.

Under Input Generator, select PDB Reader Manipulator. Scroll down to the input box. Ensure the PDB option is selected before inputting 4HCO for PLK1.

On the next page, deselect protein chain B to only process chain A.Proceed. Scroll down the PDB Info and proceed. Of the four outputs, we select the PDB file and save it.

Finally, proceed to open the PDB file with a text editor. Find and replace the histidine protonation state from HSD to HIS. Natural product database screening.

The approach to obtaining a sample of small molecules to target PLK1 employs a natural products database, and has three parts. Natural product databases are extensive, so initial compounds are from selecting a disease and confidence limit. In this example, SuperNatural 3.0 is used.

SuperNatural 3.0 is used to obtain molecules through disease selection. One way to query is through human pathways. Here breast cancer is selected.

The KEGG ID for this pathway is here. Then we type the KEGG ID into SuperNatural 3.0 pathway. Alternatively, we can click Disease, select breast cancer, and save the complete result file.

In frequently asked questions, the entire library of the database can be downloaded.Proceed. Proceed to unzip the zipped files, obtain the CSV, and remove the zip file. Open R Studio, or an IDE of your choice.

After importing data in R Studio, the initial complete results have 73, 046 molecules. There are 1, 193 molecules after application of confidence limits. Cluster sampling.

After applying Lipinski's rule of five with the RDkit. Chem package, we are left with 999 molecules. Here it is evident that we have 999 molecules.

Next, we apply K-means clustering to derive 50 representative molecules. Clustering is based on Tanimoto similarity. Protein-ligand docking and binding affinity calculation.

The goal of this method is to dock PLK1 to each ligand for binding affinity. Proceed to open CB-Dock2. Proceed to the Dock tab.

Here we upload the 4hco. pdb as the target protein for docking. Upload a ligand.

Here we start with ligand one. Alternatively, you can type or paste in this SMILES string. Proceed to type in your email for ease of data collection.

Afterward, click Auto Blind Docking. There are results sent in the email. Save the zip file of the docking results.

After saving, the zip file is extracted, and within it are a variety of files. Remove all files, with the exception of the complex. pdb files.

There should be a default five complex. pdb files. Clean the directory by removing the unextracted zip files.

After completion of protein-ligand docking with CB-Dock2, which ranked poses by Vina scores, The next step is calculation of binding affinity with PRODIGY. Binding affinity with PRODIGY ensures competitive assessment of protein selectivity. It is favorable for ligands to be selective of PLK1, and not the structural homologs.

Proceed to the PRODIGY web server. Select the PRODIGY-LIGAND tab. Refer to a complex.

pdb file to determine the protein and ligand chain identifiers. The complex. pdb file is opened with a text editor.

The chain identifier for the protein is P.That is inserted into the web server. The ligand chain typically starts at HETATM. Search for that.

The chain identifier for the ligand is A:UNL. That is inserted in the web server. The five complexes provided for each protein ligand combination will be uploaded at once.

A developed file management script is used in R for this task. The script zips the pdb file for all docking combinations. After zipping, the files are manually uploaded to PRODIGY.

Proceed to submit PRODIGY ligand. There's a list of binding affinities for each five poses in the uploaded combination. Delta-Gnoelec is the binding affinity score excluding electrostatic calculations.

Proceed to save the results, which are zipped. Upon opening the zip file, there is an output folder with the affinity scores. ADMET-S evaluation.

The ADMET evaluation, without stability, involves three different tools. ADMET assesses druggability and bioavailability using over 120 properties. ORCA applies the Density Functional Theory, DFT, to analyze drug properties.

Let's proceed. Create a directory for stability. Retrieve the SMILES string of a small molecule.

Open Avogadro, or molecule editor of your choice. Use the SMILES string to build your small molecule. Save the molecule after building.

Ensure it is located in a stability subdirectory. Proceed to Optimize Geometry. Next, an ORCA input file is generated.

Copy and paste the desired level of theory, such as seen from the fourth line of the example document. Proceed to replace the default level of theory. Proceed to generate the file.

Save in this stability folder. Add the sh file or bash job file. Proceed to customize parameters of the job file.

For example, the email to be notified can be customized. Representative results. First, we have PCA for multiple dimensionality reduction.

This figure shows different numbered clusters grouped in gray ellipses. Next, results of sample ligand 1 and PLK1-PBD. On the top-left is a portrayal of ligand 1 in green, with undesired residues, red and orange.

On the right is a representative model after cleaning and preparation with CHARMM-GUI. Below is the docked complex with PLK1, PBD, and ligand 1. Additionally, CB-Dock2 demonstrates the top five docking cavities and their affinities in the form of Vina scores.

Predicted binding affinities from PRODIGY can be visualized with heatmaps. The heatmap portrays roles of homologs PLK1, 2, and 3. Red represents weak, undesirable affinity, and green the opposite.

ADMET-S evaluation can be represented with a variety of figures. First is the boiled egg. The heatmap identifying inhibitor and non-substrate status with green and red of ligands against different metabolic indicators.

For clearance, a line graph portraying ideal time thresholds for the drug to be distributed before expression. Various metrics can also be compiled into a useful radar chart. The number of toxicophores in each ligand can be portrayed with a column chart.

Toxtree outputs both a drug toxicity classification and qualitative reasoning based on Cramer's rule decision tree. Stability or molecular reactivity can be portrayed with a line graph of adequate HOMO-LUMO gap thresholds.Conclusion. To summarize, this protocol uses a variety of virtual drug screening, binding affinity, and ADMET-S tools at the high school or undergraduate level.

The results generated from this protocol has potential to increase the yield of favorable drug discovery.