Generative Therapeutic Design (GTD) is a cutting-edge solution for exploring chemical space starting from a set of initial molecules and target properties to generate highly plausible molecular candidates see our recent publication with Gilead.
Through iterative cycles of molecular generation and pruning, GTD refines these candidates, but the evolutionary process can be opaque and complex.
Understanding Chemical Space
To support a better understanding of chemical space explored in an experiment and explain where new molecular candidates originated, a new visualization feature has been developed. This feature, called the Pedigree Tree, provides an interactive, graphical representation of a molecule’s transformation history. Built using Angular and D3.js, it aims to improve both transparency and interpretability of the molecular design process.
This component is integrated as a dedicated visualization tab within each compound’s detailed view. When a molecule is selected, the tree is dynamically generated and centered on that molecule, allowing users to trace its origins and view related molecules along its lineage. By clicking on a node, users can access relevant molecular information such as structural representation, transformation details, property values, and performance scores.
The visualization also includes a color-coded desirability scale, offering an immediate sense of how well each molecule performed against a given optimization criterion.
Pedigree Tree Visualization
Displayed as a tree diagram, the Pedigree Tree shows how a molecule has been arrived at through multiple transformation steps, starting from a user supplied input compound. Each node in the tree represents a molecule, and each edge corresponds to an alteration made by the GTD generative engine that produces new candidates from previous ones. Proceeding from left to right in the visualization, input molecules are represented as root nodes, and the branches illustrate the various paths that GTD has explored to reach a final structure shown at the right-hand edge of the diagram.
Beyond simply showing how a molecule was generated, the Pedigree Tree offers a broader view of the entire exploration path. It helps identify not only promising design directions, but also less successful branches, cases where no well-rated molecules were produced. This visibility helps identify where filter settings or applicability domain may be cutting off exploration. It also clarifies which input molecule classes have been thoroughly explored and which were quickly outcompeted.
In summary, the Pedigree Tree is a powerful tool for anyone seeking to understand the molecular evolution process in GTD. It brings clarity to complex experiment results and supports more informed design and parametrization of subsequent GTD experiments.
📩Want to find out the latest news about BIOVIA events, customer stories, blogs and more?
Damien LegrandDamien Legrand is currently pursuing a Master’s degree in Computer Science with a focus on Data / Artificial Intelligence at Institute Polytechnique de Paris. He recently completed a 5-month internship with the Bioscience Software Development team at BIOVIA headquarters in San Diego California. Outside of work he is an avid player of basketball and other sports and bring a competitive spirit both on the court and when tackling problems at work.
Leo BleicherLeo Bleicher is the Director of Bioscience and Scientific Informatics Development at BIOVIA in San Diego. He spent two decades working as a medicinal chemist, cheminformatics and drug discovery information scientist before joining BIOVIA in 2012