• ES Español
• EN English

3.14. Computational and Theoretical Chemistry (CTC)

Computational and Theoretical Chemistry provides theoretical foundations and computational tools for predicting and simulating chemical behavior. This computing-focused subset covers molecular modeling and simulation, machine learning for chemical property prediction, and AI-driven drug design, directly relevant to bioinformatics, AI, and data science.

Knowledge Area (KA)	CS Core	KA Core
3.14.1 Molecular Modeling and Simulation	Elective
3.14.2 Machine Learning in Chemical Discovery	Elective
3.14.3 Computational Drug Design	Elective

3.14.1. CTC/Molecular Modeling and Simulation ↑ Back to top

Principles of molecular mechanics and dynamics simulations for studying molecular structure and motion, with connections to physics-based simulation and sampling methods in computing.
Topics:
Core

• Force field potential energy functions: bond, angle, torsion, and non-bonded interaction terms
• Molecular dynamics: numerical integration algorithms, periodic boundary conditions, and thermostats
• Monte Carlo methods for conformational sampling: Metropolis criterion and importance sampling
• Implicit and explicit solvation models for condensed-phase molecular simulations
• Free energy perturbation and thermodynamic integration for relative binding affinity estimation

Learning Outcomes:
Core:

1. Describe the components of a standard molecular mechanics force field and the physical interpretation of each energy term [Familiarity]
2. Set up and run a molecular dynamics simulation, monitoring energy conservation, temperature equilibration, and structural observables [Usage]
3. Evaluate the stability and convergence of a molecular dynamics trajectory using RMSD, energy profiles, and thermodynamic observables [Assessment]

3.14.2. CTC/Machine Learning in Chemical Discovery ↑ Back to top

Application of statistical learning and deep learning to predict molecular properties, accelerate chemical discovery, and generate novel molecular structures relevant to AI and data science.
Topics:
Core

• Molecular representations: SMILES notation, molecular fingerprints, and molecular graph encodings
• Quantitative structure-activity relationship (QSAR) models using supervised regression and classification
• Graph Neural Networks (GNN) for learning molecular properties directly from chemical graph structures
• Generative models (VAE, diffusion, transformer-based) for de novo molecular design
• Model validation: cross-validation, applicability domain, and chemical space diversity in training sets

Learning Outcomes:
Core:

1. Define and compare different molecular representations and explain their suitability for specific machine learning tasks [Familiarity]
2. Train a QSAR regression model to predict a molecular property from descriptors and evaluate its performance using standard regression metrics [Usage]
3. Evaluate the validity, diversity, and novelty of molecules generated by a deep learning model using standard benchmarking metrics [Assessment]

3.14.3. CTC/Computational Drug Design ↑ Back to top

Structure-based and ligand-based computational approaches for identifying and optimizing candidate drug molecules, relevant to AI-driven drug discovery pipelines.
Topics:
Core

• Molecular docking: conformational search algorithms and scoring functions for binding pose prediction
• Virtual screening of compound libraries against protein targets to prioritize drug candidates
• ADMET property prediction: absorption, distribution, metabolism, excretion, and toxicity profiles
• Pharmacophore modeling and ligand-based virtual screening approaches
• Protein binding site identification and characterization for structure-based drug design

Learning Outcomes:
Core:

1. Identify key intermolecular interactions between a ligand and a protein active site from a computationally docked complex [Familiarity]
2. Perform a molecular docking study, rank predicted binding poses by scoring function values, and select candidates for follow-up analysis [Usage]
3. Design and execute a virtual screening workflow to identify potential enzyme inhibitors from a compound database [Assessment]