The EU-funded RISK-HUNT3R project is researching the development of the next generation of methods for the animal-free risk assessment of new substances. The aim is to use computer-assisted methods to fully assess the toxicological and ecological risks of new chemicals without having to synthesize the chemical compounds and test them in cells and animals.

However, regulatory authorities are sceptical because they cannot understand the computers' deep learning systems and they are not transparent, meaning that any errors cannot be detected.

Graphic: Computer-generated.

The Viennese scientists have now developed the software tool “MolCompass”, which is designed to detect weaknesses in prediction tools that establish a quantitative relationship between the pharmacological, chemical, biological or physical effect of a molecule and its chemical structure (QSAR/QSPR models).

For this purpose, the Viennese researchers developed interactive graphical tools that project chemical compounds onto a 2D level. Colors are then used to highlight compounds that were predicted incorrectly with a high degree of certainty. These can then be investigated separately using other methods.

Original publication:
Sosnin, S. MolCompass (2024). multi-tool for the navigation in chemical space and visual validation of QSAR/QSPR models. J Cheminform 16, 98. https://doi.org/10.1186/s13321-024-00888-z

Further information:
https://extrajournal.net/2024/08/21/ai-soll-tierversuche-ersetzen-macht-aber-fehler-tool-der-uni-wien-hilft/