Praexisio, an integrated bioinformatics and preclinical consulting services company founded by Professor Lee-Wei Yang, director of the PhD program in Intelligent Biomedical Science at National Tsing Hua University, is utilizing generative AI combined with molecular dynamics simulations to redesign old drug compounds into new medications. By screening compound fragments from a global database of FDA-approved drugs and using computer-assisted technology to reassemble them into new drugs, Praexisio's approach not only shortens drug development timelines but also improves the efficacy and safety of new drugs, thereby increasing the overall success rate of drug development.

The innovative technology allows for significant reductions in the time required for drug discovery and preclinical development, compressing a process that traditionally takes over five years into less than two years. Moreover, a new drug for triple-negative breast cancer developed using this technology has already shown preliminary efficacy and safety in mouse models, demonstrating strong market potential.

Empowering Drug Design with Protein Molecular Dynamics Simulations and Generative AI

High failure rates in drug development have long been a major challenge for the pharmaceutical industry. Developing a new drug is time-consuming, labor-intensive, and costly. Even after successful clinical trials, safety issues such as side effects can still arise post-approval. Praexisio addresses these issues by integrating generative AI with protein molecular dynamics simulation technology, providing a novel approach to small-molecule drug design that reduces the likelihood of these problems.

In Praexisio's design process, an AI system is trained to screen compound fragments from FDA-approved drugs, which are considered relatively safe. These fragments are then recombined through computer-assisted techniques to design new drugs. This method effectively mitigates potential efficacy or systemic performance issues that often cause drug candidates to fail during late-stage clinical trials, thus improving the chances of successful drug development.

Praexisio has acquired relevant drug molecule design and measurement technologies through a technology transfer and licensing agreement with National Tsing Hua University. These technologies are based on nearly half a century of research on the dynamic conformation of proteins. The core of the technology involves selecting several conformations of a target protein during the screening stage, followed by simulation docking using a small-molecule database. The AI system, trained on FDA-approved drugs, then reorders these docking results, integrating molecular dynamics simulations under physiological conditions to identify the optimal "on-target compound fragments."

Based on these findings, Praexisio conducts biochemical and cancer cell inhibition experiments on the top 5 to 10 compounds, analyzing the results in conjunction with previous simulations to determine which compound fragments from FDA-approved drugs exhibit potential "systemic benefit."

Innovative Drug Recombination with Generative AI

Once potential compound fragments are identified, Praexisio further utilizes generative AI to reassemble these fragments and synthesize new drugs. This process involves iterative simulations, biochemical experiments, cell-based assays, and animal studies to confirm the efficacy and safety of the new drugs. Several triple-negative breast cancer drugs designed using this strategy have already demonstrated Namur-level molecular inhibition effects and dose-dependent anti-cancer effects with low toxicity in mouse models, showcasing promising application prospects.

In addition to addressing efficacy issues in drug design, Praexisio incorporates gene expression data into its drug screening process. When a disease causes changes in the expression of multiple genes, the therapeutic potential of a drug can be inferred if its interference effects are highly negatively correlated with those gene expression changes. This gene-expression-based screening method significantly enhances the accuracy and effectiveness of drug development.

Generative AI plays a critical role in Praexisio's drug design process. Professor Lee-Wei Yang refers to this as "disciplined generative AI," where the system does not randomly generate entirely new and unknown compounds but operates within a controlled latent space that combines targeted inhibition with systemic benefits. This ensures that the generated compounds are safe and effective without introducing unknown risks.

Systemic Considerations and Automated Optimization in Drug Design

Through collaboration with the PhD program in Intelligent Biomedical Science at National Tsing Hua University, Professor Ivet Bahar of the National Academy of Sciences, and the University of Alberta in Canada, Praexisio’s innovation extends beyond generative AI. From the early stages of drug design, systemic benefits are considered to ensure that the designed drugs not only have high affinity for the target protein but also exhibit overall efficacy. This systemic approach enhances the therapeutic effect of the drug.

Praexisio’s automated pipeline continues to optimize the efficacy, bioavailability, half-life, toxicity, and other ADMET (Absorption, Distribution, Metabolism, Excretion, and Toxicity) parameters of the drugs in the latent space. Additionally, since many of the compound fragments are derived from FDA-approved drugs, they have clear references for large-scale synthesis, making manufacturing issues less likely and ensuring higher predictability and stability. This greatly reduces uncertainty in production.

Professor Lee-Wei Yang further explained that while generative AI is a key component of Praexisio’s drug development platform, it is not the only one. "We aim to integrate supervised AI, generative AI, and first-principles simulation to solve current challenges in drug development and to reassess these challenges from a new perspective."

Future Prospects

Praexisio’s drug design platform, which integrates generative AI, molecular dynamics simulations, and preclinical research, offers new possibilities for the pharmaceutical industry. Through collaboration with top research teams both domestically and internationally, the company has successfully shortened drug development cycles and improved success rates. As AI technology continues to advance, Praexisio’s innovative platform is poised to drive further progress in drug design, providing the pharmaceutical industry with more promising solutions.

Resource (mandarin): Praexisio重組老藥片段設計新藥 縮短時程還能提升成功率