A new artificial intelligence model developed by researchers in Sweden could significantly speed up the process of testing new drugs, according to findings presented in a study published in Science Advances on June 11. The technology is expected to help identify promising drug candidates more quickly and with greater accuracy.
Drug development typically takes over ten years from conception to final product, with much of the cost and time concentrated in early-stage testing. Thousands of molecules are screened during these stages, but only a small fraction advance for further development. Traditionally, simulating molecular movements has relied on molecular dynamics calculations that require billions of computational steps due to the need for extremely short time increments.
The new AI approach allows researchers to detect molecular changes without extensive numerical calculations. Machine learning accelerates each calculation step, while generative models can generate plausible molecular structures directly. A team from Chalmers University of Technology and the University of Gothenburg has created an AI model that is reported to be more than 10,000 times faster than conventional simulations.
The study involved analysis of over 12,500 organic molecules and more than a thousand short peptides. According to Simon Olsson, "We train the model using simulated examples of how the atoms in a molecule move over time. Based on these sequences, the model learns the underlying rules governing the movement of the molecules and can then predict how new molecules will behave." Olsson added that results were validated using post-processing simulations: "We validated the results using extensive post-processing simulations to corroborate them using standard numerical algorithms, and they are consistent with one another." He described this as akin to jumping between scenes in 'molecular movies' rather than watching every frame sequentially.
The AI model reportedly generalizes well even for previously unseen molecules because it learns general rules rather than memorizing specific systems. Olsson said, "There is a certain pattern that the model helps us to identify... So, with the help of artificial intelligence, we can work out what is likely to happen in the `molecular future`. It can predict how molecules change even though it has never seen the process unfold." Juan Viguera Diez stated, "In order to be able to predict the physical phenomena exhibited by molecules, we need to understand the underlying physics of how the system behaves. I believe we are among the first to demonstrate this in a general sense and show that it is possible."
Researchers hope this advancement will contribute toward more efficient drug development by enabling rapid simulation screening at early stages. Viguera Diez concluded, "In the long term, AI models like ours could help to identify promising drug candidates more quickly and improve accuracy in early stages... This will hopefully pave way for development of more general techniques which may ultimately facilitate development of new drugs and treatments—and also improve our understanding of diseases." The TITO (Transferable Implicit Transfer Operators) framework forms the basis for these predictions but so far has been tested mainly on small systems under simplified conditions.
