Computational Biology

(BIOSC 1540)

Oct 31, 2024

Lecture 17:
Docking and virtual screening

Announcements

• A06 is due tonight by 11:59 pm
  • Reminder: There is a (soft) limit of 100 words for each question
• A07 (final assignment) will be released tomorrow
• No class on Tuesday (Nov 5) for election day
• The next exam is on Nov 14
  • We will have a review session on Nov 12
  • Request DRS accommodations if needed
• Project will be released Nov 21 and is due ???

After today, you should have a better understanding of

Need limitations of alchemical simulations

Alchemical simulations: Precise but expensive

Why are alchemical simulations computationally expensive?

Efficient methods are needed for virtual screening

Remember: Chemical space is unfathomably large and the role of computation is to virtually test as many compounds as possible

Value of data-driven approaches for drug discovery

After today, you should have a better understanding of

Accurate and efficient binding predictions are essential

Docking simplifies the binding free energy prediction problem to enhance speed

Identifying relevant protein conformations

After today, you should have a better understanding of

Accurate, reproducible docking requires a relevant protein conformation

Challenge: Proteins Are Dynamic Molecules

Sources of Protein Conformational Data

Experimental Structure Selection Criteria

• Resolution and Quality
  • Prefer structures with higher resolution (e.g., <2.5 Å).
  • Assess reliability using R-factors and validation reports.
• Ligand-Bound vs. Apo Structures
  • Ligand-Bound (Holo) Structures
    • Provide direct insight into binding site conformation.
  • Apo Structures
    • May reveal binding site flexibility in the absence of ligands.
• Relevance to Target Ligand
  • Choose structures co-crystallized with ligands similar to those of interest.

Molecular Dynamics Simulations for Conformational Sampling

Extract representative structures using clustering algorithms.

Identify conformations with open or closed binding sites.

Importance of Water Molecules

Detecting binding pockets

After today, you should have a better understanding of

Accurate Binding Pocket Detection is Crucial for Docking

The binding pocket is the specific region where a ligand interacts with a protein

Accurate identification of binding pockets is essential for successful docking and virtual screening.

Understanding Protein Surface Topography

Classification of Binding Pockets

• Orthosteric Sites: The primary active site where endogenous ligands bind.
• Allosteric Sites: Secondary sites that modulate protein function upon ligand binding.

• Cryptic Sites: Binding pockets not apparent in the unbound protein structure but form upon ligand binding or conformational change.

Geometry-Based Pocket Detection Techniques

Alpha Shape Theory: Uses Delaunay triangulation and alpha complexes to define cavities.

Grid-Based Pocket Detection

• Methodology
  • Overlay a 3D grid on the protein structure.
  • Classify grid points as inside, outside, or on the surface.
• Pocket Identification
  • Clusters of surface grid points forming concave regions indicate potential pockets.

Detecting Cryptic Binding Sites

Ligand pose optimization

After today, you should have a better understanding of

Accurate Docking Depends on Optimized Ligand Poses

• Precise ligand poses are crucial for reliable predictions of binding affinity and activity.
• Incorrect poses can lead to false negatives or positives, misguiding drug development efforts.

Fundamentals of Ligand Pose Optimization

• Definition of Ligand Pose
  • The specific orientation and conformation of a ligand within the binding site of a target protein.
• Optimization Goal
  • Identify the energetically most favorable pose that closely represents the true binding mode.
• Key Components
  • Orientation: Position and alignment within the binding pocket.
  • Conformation: Internal geometry, including bond angles, lengths, and torsions.

Docking needs to generate diverse conformations

Search strategies

• Systematic
• Stochastic
• Empirical
• Machine learning

Systematic searches numerically iterate over all possible conformations

Identify important degrees of freedom

• Angles
• Dihedrals

Scan along each angle with a step size of a N degrees

Remove structures with high strain

Systematic searches are only possible for very small molecules

How many different conformations would we have in this molecule if we scanned only dihedrals every 45 degrees?

Systematic searches are only possible for very small molecules

8 dihedrals

1

2

3

4

5

6

7

8

8 angles

8 × 8 × 8 × 8 × 8 × 8 × 8 × 8 = 16,777,216

That's a lot of structures, and many of them will clash!

We almost never do a systematic search in practice without some precautions to combinatorics

Stochastic algorithms provide better balance of sampling and cost

We also have conformer libraries

Use pre-generated libraries

Scoring functions as data-driven predictors

After today, you should have a better understanding of

Scoring function are parameterized models to estimate binding affinity after docking

Interpretation of docking results

After today, you should have a better understanding of

Case study: MAP kinase ERK2

Overview of ERK2:

• Extracellular signal-regulated kinase 2 (ERK2), encoded by the MAPK1 gene, is a crucial component of the mitogen-activated protein kinase (MAPK) signaling pathway.
• ERK2 regulates vital cellular processes, including proliferation, differentiation, and survival.

Role in Disease:

• Aberrant ERK2 activity is implicated in various cancers, often due to mutations in upstream components like RAS and RAF, leading to uncontrolled cell growth.
• ERK2 is also involved in inflammatory diseases and neurodegenerative disorders.

AlphaFold 2 Prediction

MolModa

Evaluating and Scoring Binding Pockets

• Volume and Depth: Larger, deeper pockets may accommodate larger ligands.
• Hydrophobicity/Hydrophilicity: Influences ligand binding characteristics.
• Amino Acid Composition: Presence of key residues for interactions.

Challenges and Limitations

• Protein Flexibility
  • Static structures may miss important conformations.
• Water Molecules
  • Role of water in the binding site can complicate detection.
• Computational Resources
  • High-resolution methods may be computationally intensive.
• False Positives/Negatives
  • Overprediction of non-druggable sites or missing true sites.

Accurate and efficient binding predictions are essential

Thermal shift assays

High-Quality Datasets Enable Better Predictions

• Utilizing Crystallographic Structures
  • Real protein conformations guide accurate docking.
  • Reveal detailed information about binding sites and interactions.
• Incorporating Experimental Affinities
  • Adjust predictions based on known binding strengths.
  • Calibrate computational methods using empirical measurements.

Empirical Methods Improve Docking Accuracy

• Utilizing Crystallographic Structures
  • Real protein conformations guide accurate docking.
  • Reveal detailed information about binding sites and interactions.
• Incorporating Experimental Affinities
  • Adjust predictions based on known binding strengths.
  • Calibrate computational methods using empirical measurements.

Empirical Data Reduces Computational Effort

• Efficiency Gains
  • Focus on experimentally validated binding modes.
  • Limit the search space to plausible interactions.
• Resource Optimization
  • Achieve accurate results without extensive computations.
  • Save time and computational power for other tasks.

Empirical Data Increases Confidence in Results

• Validation of Predictions
  • Compare computational outcomes with experimental data.
  • Confirm the accuracy of docking models.
• Enhanced Decision-Making
  • Reliable predictions support better drug discovery choices.
  • Reduce the risk of pursuing ineffective compounds.

Introduction to Data-Driven Approaches

Data-driven approaches leverage large datasets and machine learning algorithms to make predictions, identify patterns, and generate models without relying solely on physical or chemical principles.

These methods harness the power of available biological data to accelerate research and discovery processes.

Advantages of Data-Driven Approaches

• Speed and Efficiency: Capable of processing and analyzing large datasets rapidly, facilitating quicker decision-making.
• Scalability: Easily scalable to handle growing amounts of data without a significant increase in computational resources.
• Flexibility: Adaptable to various types of data and applications, from genomics to drug discovery.
• Cost-Effectiveness: Reduces the need for expensive experiments by predicting outcomes computationally.