Lecture 10 - Docking and Drug Discovery Flashcards
What are the two main types of pharmaceuticals?
1.Biologics: Large, complex molecules produced using living organisms like bacteria, yeast, or mammalian cells. For example, monoclonal antibodies, insulin, growth hormones, and vaccines.
2.Small Molecules: “Traditional” pharmaceuticals chemically synthesized in laboratories through a series of chemical reactions. They are typically smaller than biologics and cheaper to produce. Examples include aspirin and paracetamol.
How are drug targets found and defined?
Drug targets are found and defined through experimental studies to define disease mechanisms.
Bioinformatics plays a crucial role in target identification through omics data and network data utilization.
What are ‘new’ and ‘established drug targets?
Established drug targets have a good scientific understanding of normal and pathological physiology and benefit from drug industry experience.
New drug targets include newly discovered proteins and proteins with newly discovered function / disease involvement.
What are some strategies to drug a target?
1.Design a molecule specific to this target “from scratch” (de novo drug design).
2.Find out in nature what molecules can recognize and inhibit/activate this target.
3.Go through a large “library” of molecules and see which one(s) work(s) (Drug screening).
In practice, a combination/blend of these strategies is often used, with optimization & customization of the drug “hit” being crucial.
What is the ChEMBL database, and what are some other examples?
A manually curated chemical database of bioactive molecules with drug inducing properties. It is maintained by the European Bioinformatics Institute (EBI).
Other examples include PubChem, DrugBank, and FDA collections.
What are high affinity and potency in the context of pharmacokinetics, and how can they be measured?
High affinity shows good binding to the drug target. The Kd is a good measure of affinity (lower Kd, higher affinity)
High potency shows effective and efficient in its desired activity. The Half maximal inhibitory concentration (IC50) is a measure of potency. It quantifies the concentration of a drug which inhibits the target by 50%
What molecular weight is classified as a small molecule?
A “small molecule” is an umbrella term for organic molecules of low molecular weight, typically below 1000 Da.
How are the structures of small molecules represented?
The structures of small molecules can be represented using formats like InChI, InChI Key, SMILES, and Molecular formula.
What are the requirements for Drug Screening?
Structure of the Drug (essential)
Sequence and structure of the target
If the target’s properties are not available, we can perform ‘ligand-based screening’.
What is structure based drug screening and design?
Where the structure drives assumptions when designing a drug.
Here, it is assumed the specificity & activity of the drug is determined by the complementarity between the structure of the drug and the shape of the drug-binding site on the target.
Drug-binding sites need to be accessible, i.e., exposed to solvent.
What is in silico drug screening or virtual docking?1.
In silico drug screening or virtual docking aims to compute “poses” of drugs bound to the target and repeat this for a large library of drugs.
A pose refers to the structural coordinates with the protein bound with a given ligand (drug).
What is a virtual screening workflow?
- Library prep (Obtaining of compounds)
- Structure and Ligand-based filtering
- Experimental Validation (In-vivo and In-vitro assays)
What is fragment based screening?
Involves testing libraries of chemical fragments instead of fully-elaborated ligands.
A larger ligand with better affinity and activity can then be grown from these fragments.
While this helps with the computational cost upfront and provides a better chance of obtaining hits, this then requires significant efforts post-screening to obtain the final ligand.
What are the two main phases of Molecular Docking?
1.Placing molecules into the binding site of the target (pose identification).
2.Forecasting the strength of binding between the docked conformation and the target (scoring).
What is the main hindrance in molecular docking?
The shortcomings of existing scoring functions persist as a significant hindrance for the molecular docking process.
What are the energetics of protein docking?
Prediction of the correct structure (pose) of the bound complex does not require knowing the Ka. However, prediction of biological activity does.
Important energetic factors to consider are: Steric, electrostatic, hydrogen bonding, strain, desolvation, and rotational and translational entropy.
What are some of the molecular difficulties in docking?
Because both molecules are flexible, there are an astronomical number of complex combinations.
The solution used for this problem is to treat the protein as rigid, and restrict the flexibility of the molecules.
This includes various methods including: rigid ligand and target, or flexible ligand and rigid target.
What are the three main approaches to scoring functions in molecular docking?
Force-field approach
Empirical approach
Knowledge-based approach
Describe the force-field approach to scoring functions.
Considers physics-based terms like Coulombic interactions (electrostatics) and the Lennard-Jones model (van der Waals’ interactions).
Describe the empirical approach to scoring functions.
Based on first principles, this approach takes into account ground truths such as Van-der Walls forces, hydrogen bonding, and the hydrophobic effect.
It then builds linear regression fit to experimentally determined binding affinities to produce scoring coefficients.
Describe the knowledge-based approach to scoring functions.
Based on knowledge obtained through various experiments, for example, through potentials of mean force (PMF) fitted over defined sets of experimentally determined structures.
How are docking programs evaluated?
Decoy” ligands are typically used as negative controls to evaluate how well a docking program has performed.
A good docking method should have known binding ligands enriched in the hit list over decoys.
What is the benefit of combining multiple docking methods?
Combining multiple docking methods can help to narrow down the hit list and consensus between methods, potentially leading to more reliable predictions.
What is a database used for drug screening?
ZINC20
What is AlphaFold3 and what new capability does it have compared to AlphaFold2?
AlphaFold3 is an improvement over AlphaFold2 with new support for predicting protein in complex with other proteins, DNA/RNA, and small molecules using a SMILES string as input.
Implementation differences include a decreased reliance on multiple sequence alignment (MSA) and the use of a diffusion module for structure prediction.
Why is Antibody-Antigen Docking a challenging case?
Antibody-antigen docking is a challenging case due to the structural diversity of antibodies and the variability of antigens.
Additionally, limited and noisy experimental data further complicate the training and validation of AI/ML models, requiring specialized methods.
