New open-source tool helps researchers explore complex chemical data
Researchers have developed a new open-source workflow that transforms large chemical datasets into interactive visual maps, making it easier to identify patterns in chemical structures and properties. The work, supported by Mistra SafeChem, could help researchers make better use of data in the development of safer and more sustainable chemicals.
Modern chemistry generates vast amounts of data, yet extracting useful insights from thousands of chemical structures remains a major challenge. Researchers often need to compare large numbers of compounds simultaneously to understand how chemical structure influences properties such as biological activity, toxicity, or environmental behaviour.
In a recent study published in Molecular Informatics, researchers present an automated workflow that helps address this challenge. Developed within the open-source KNIME Analytics Platform, the workflow creates interactive visualisations known as Constellation Plots, allowing users to explore relationships between chemicals and their properties in an intuitive way.
"As chemical datasets become larger and more complex, there is a growing need for tools that help researchers identify meaningful patterns and make informed decisions. Our goal was to make advanced chemical space analysis more accessible to a broader community of scientists," says Edgar López, researcher at Uppsala University, co-author of the study and member of the Mistra SafeChem programme.
The visualisations group chemicals according to shared core structures and display them in an interactive map. Researchers can quickly identify families of related compounds, compare their properties and investigate how structural changes influence performance or activity.
Instead of examining compounds one by one, users can gain an overview of an entire chemical landscape in a single visualisation. The maps can also be explored interactively, allowing researchers to click on specific groups of chemicals and immediately access the underlying data.
Supporting data-driven approaches to safer chemical design
To demonstrate the workflow, the researchers analysed a dataset containing more than 5,200 compounds that inhibit the Tau protein, a therapeutic target associated with Alzheimer's disease. The study identified more than 3,200 unique structural groups, illustrating the diversity that can exist within a single chemical dataset.
Although the case study focused on drug discovery, the workflow can be applied to virtually any chemical dataset containing structural information and measured or predicted properties.
For researchers working with Safe and Sustainable by Design approaches, tools that help visualise links between chemical structure and substance properties are becoming increasingly important. Such insights can help scientists navigate large datasets, identify promising chemical candidates and better understand patterns associated with desirable or undesirable characteristics.
"While we demonstrated the workflow using pharmaceutical data, the approach has much broader applications. It can support researchers working across different areas of chemistry, including efforts to develop chemicals and materials that are safe and sustainable by design," says Edgar López.
The workflow is freely available through GitHub and is built entirely using open-source software, supporting transparency, reproducibility and wider access to advanced chemoinformatics tools.
As the volume of chemical data continues to grow, visualisation tools such as these may play an increasingly important role in helping researchers turn complex datasets into actionable knowledge.