Aiscra Project

What problem do we want to solve?

Molecular docking is a commonly used virtual screening method in the pharmaceutical industry, guiding initial decisions in drug design. However, this technique does not definitively determine a compound's activity, and its results depend heavily on the accurate execution of the docking procedure for the chosen biological target and the proper interpretation of the predicted ligand poses. The process of analyzing molecular docking results and selecting the most promising compounds for further studies is tedious and requires expertise in various fields, such as medicinal and computational chemistry, as well as proficiency with specialized computer software.

Our solution

We propose a platform called Aiscra, which aims to simplify and automate key steps from conducting molecular docking to selecting the most promising candidates for laboratory testing. Interaction with the platform will occur through conversations with a virtual assistant based on large language models. This will enable easy configuration of individual cheminformatics tools using natural language commands. The platform will include tools to analyze the quality of ligand poses, predict crucial molecular properties for early drug design stages, filter predicted poses based on preferred protein interactions, and even automatically generate analogs of analyzed compounds. The Aiscra platform will help analyze and select compounds to increase the likelihood of finding hits and reduce the risk of failure in later stages of drug design. The functionality of the platform will be tested in the project by performing biochemical assays and crystallographic experiments.

Contact Us

If you have any questions about our project or want to help build or test our solution, please send us a message at tomasz.danel@uj.edu.pl.

Acknowledgments

Project funded by the National Center for Research and Development under the LIDER XV program.