Drug Docking and Design Flashcards
Week 7 Lecture 1
What are the types of drugs?
- Small chemical molecules (e.g. aspirin)
- Simple biologics (e.g. insulin)
- Larger biologics (e.g. somatotropin)
- Complex biologics (e.g. monoclonal antibody)
What are biopharmaceuticals?
Biopharmaceuticals (biologics) are large, complex molecules produced using living organisms. These living cells, such as bacteria, yeast, or mammalian cells serve as factories where the biopharmaceuticals are synthesized through biological processes.
What are traditional pharmaceuticals?
Traditional pharmaceuticals are small molecules that are chemically synthesized in laboratories through a series of chemical interactions.
Differences between biopharmaceuticals and traditional pharmaceuticals
- Complexity: Biopharmaceuticals are more complex because they are made of proteins, nucleic acids, or other large molecules.
- Specificity: Biopharmaceuticals are often highly specific, targeting specific proteins or cells in the body.
- Regulation: Biopharmaceuticals are more highly regulated due to their complexity and their involvement in living organisms. They are regulated for their safety, efficacy, and consistency.
- Cost: Biopharmaceuticals are more expensive to produce due to their complexity and the need for specialised facilities.
What is a drug target?
A natural molecular structure involved in the pathology of interest. Can be proteins (receptors, ion channels, enzymes), DNA, RNA and ribosomal targets.
What is a lead compound?
A chemical compound or natural product that exhibits some level of biological activity against a specific drug target. This activity serves as a starting point in the drug discovery process.
e.g. aspirin, morphine, penicillin
What is druggability?
- How well a therapeutic can access a target.
- The efficacy depends on whether the therapeutic can achieve effectiveness in modulating a specific biological target.
- Known successful target classes are GPCRs and kinases
Parameters influencing druggability
- Cellular location of target
- Development of resistance
- Transport mechanisms
- Side effects
Lipinski’s rule of 5
<= 5 H-bond donors
<= 10 H-bond acceptors
Mass <= 500 daltons
log (P) <= 5
What is ChEMBL?
A manually curated chemical database of bioactive molecules with drug-inducing properties.
What is molecular docking?
A process that helps in identifying potential drug candidates by predicting the binding affinity of small molecules to a protein or receptor of interest.
What can docking be between?
- protein/small ligand
- protein/peptide
- protein/protein
- protein/nucleotide
What are the phases of molecular docking?
- Placing molecules into the binding site of the target (pose identification)
- Forecasting the strength between the docked conformation and the target (scoring)
- The main obstacle is due to the shortcomings of the existing scoring functions
What is the docking score?
The value obtained from computational docking algorithms that predicts how well a ligand molecule fits into the binding site of a target protein. A lower docking score typically indicates a better fit or a stronger interaction between the ligand and the protein.
What is binding free energy?
A thermodynamics measure that quantifies the strength of the interaction between a ligand and a protein. It represents the difference in free energy between the bound complex and the unbound states. Lower BFE indicates a more stable complex and stronger binding affinity between the ligand and the protein.
What is binding affinity?
The strength of the interaction between the ligand and the protein. It is a measure of how tightly a ligand binds to its target protein and is often represented by the association constant (Ka), calculated as the ratio of the concentration of the complex to the product of the concentrations of the individual molecules.
Types of scoring functions in molecular docking
- Force field approach
- Empirical approach
- Knowledge-based approach
- Machine-learning approach
Force field approach
- Affinities are estimated by summing the strength of the intermolecular Van der Waals and electrostatic interactions between all atoms of the two molecules in the complex, using a force field.
- Intramolecular energies of the binding partners are often included.
- Desolvation energies in the presence of water are considered using implicit solvation methods like GBSA or PBSA.
Empirical approach
- Based on counting various types of interactions between binding partners.
- Counting may involve:
1. Ligand and receptor atoms in contact or calculating the change in solvent accessible surface area.
2. Scoring function coefficients are determined using multiple linear regression methods.
3. Interaction terms include hydrophobic-hydrophobic contacts (favourable), and hydrophobic-hydrophilic contacts (unfavourable), reflecting unmet hydrogen bonds which are an important enthalpic contribution to binding. - One lost hydrogen bond can account for 1-2 orders of magnitude in binding affinity.
Knowledge-based approach
- Relies on statistical observations from large 3D databases.
- Derives statistical potentials of mean force based on closed intermolecular interactions, assuming frequent interactions indicate favourable binding affinity.
Classical scoring functions
Force-field, empirical, and knowledge-based approaches. They assume linear combinations of their contributions to binding, limiting their ability to leverage large training datasets.
Machine-learning approach
Characterised by not assuming a predetermined functional form for the relationship between binding affinity and structural features.
- The functional form is inferred directly from data, making it more flexible.
- Outperforms classical scoring functions in diverse protein-ligand complexes and target-specific complexes.
- Particularly effective in structure-based virtual screening.
- Performance gap widens when target-specific data is available.
Difficulties in docking
- When both molecules are flexible there are hundreds of degrees of freedom and an astronomical number of possible conformations.
- Protein/ligand models often treat the protein as rigid.
Interactions between drugs and targets
- Van der Waals packing
- Electrostatics
- Hydrogen bonds
- Hydrophobic interactions
What is a pharmacophore?
A conceptual representation of molecular characteristics essential for a ligand to be recognised by a biological macromolecule.
What is a pharmacophore model?
Pharmacophore models elucidate how various structurally diverse ligands can effectively bind to a common receptor site. These models serve as tools for identifying novel ligands, either through de novo design or virtual screening, that possess the ability to bind to the same receptor.
Docking methods - flexible ligand
- Treat the receptor as static and the ligand as flexible
- Dock the ligand into the binding pocket, this generates a large number of possible orientations
- Evaluate and select by the energy function
Automated docking
There are many programs available. They all make use of:
- cavities
- surface complementarity
- electrostatics
- full energy function
How do we score docked models?
Each orientation is scored using 3 scoring schemes:
1. Shape scoring
2. Electrostatic scoring (calculates electrostatic potential)
3. Force-field scoring (uses force-field potential)
Evaluation depends on either using surface curvature/area, or grid-based evaluation of surface packing
What are decoys?
A way of evaluating how well your docking program has done on a target. Decoys refer to a set of molecules that probably won’t bind to your target.
Standard consensus
Conditional ‘AND,’ takes only the best molecules from both programs
Exponential consensus ranking
Conditional ‘OR,’ takes the best molecules from either program
Describe the virtual screening workflow.
- Library preparation - obtaining of compounds
- Filtering - structure and ligand-based filtering
- Experimental validation - in vivo and in vitro assays
Describe structure-based virtual screening
- Structure-based pharmacophore modelling
- Molecular dynamics simulation
- Molecular docking
Describe ligand-based virtual screening
- Ligand-based pharmacophore modelling
- Machine learning algorithms
- 3D shape similarity search
- Molecular fingerprints
