There are different ways to approach the task of generating new compounds using computational tools. They all have their advantages and disadvantages. Knowing the foundations of the main techniques is essential to help decide which approach is the best for a given problem.
In this series of talks Dr. James Webster walks through the fundamentals of computational generative drug design of small molecules, covering the techniques and also gives out a demonstration on how to perform it using freely available open-source software.
This was originally given as a workshop during the Drug Discovery Africa conference that happened in March 2025 in Accra, Ghana.
Workshop Materials
Slides presented at the workshop are available here: DDA CDD Workshop Slides with notes. James’ notes are available as comments and there are 5 recordings to accompany the slides.
Following this, is a series of recordings to take you step-by-step through how to use KNIME to support Computational Drug Design. KNIME nodes can be found here.
This introductory video explains the core concepts of computational generative design for small molecules, covering its purpose in drug discovery, the nature of chemical space exploration, and the fundamental pillars of molecule construction, scoring, and search.
This video covers essential cheminformatics concepts for generative design, explaining molecular representations (graphs, SMILES, InChI), reaction notations (SMIRKS), and various molecular descriptors including fingerprints and data-driven latent spaces
This video explores molecular construction in generative design, comparing atom-based, fragment-based (growing, linking, merging), and reaction-based methods along the spectrum of computational control versus synthetic accessibility.
Explore molecular scoring in generative design: this video covers explicit versus implicit scores, receptor versus ligand-based methods, balancing trade-offs with multi-objective optimization (Pareto), and applying filters to refine generated compounds.
This final video covers molecular search techniques used in generative design, explaining how methods ranging from systematic enumeration to greedy, epsilon-greedy, and simulated annealing balance exploration and exploitation to find optimal molecules in vast chemical space.
This video introduces KNIME essentials for molecule design, guiding you through installing the platform, setting up your workspace, importing workflows, customizing the interface, and using nodes to draw chemical structures.
This video explains and demonstrates atom-based, fragment-based (using bioisosteres), and reaction-based molecule construction methods within KNIME, comparing the scale, synthetic tractability, and application of the resulting molecular sets.
This video covers molecule scoring and filtering in KNIME, demonstrating how to remove duplicates and undesirable substructures, calculate key molecular descriptors, apply property rules like Lipinski’s Ro5, and export the refined dataset.
This video explains molecule search in KNIME using an epsilon-greedy strategy, demonstrating how to iteratively construct, score (via similarity to a target), and select molecules within a configurable recursive workflow to optimize structures.