TranslAI Initiative
Objective: To develop generative AI (GenAI) models that support translational research by facilitating the translation of experimental findings across domains, such as organ systems, in vitro-to-in vivo extrapolation (IVIVE), and genomic technologies.
Introduction: Translational science plays a pivotal role in various domains. This field encompasses a wide array of translational research focus with practical applications. Transitioning genomics data between platforms—such as from microarray to next-generation sequencing—offers a way to capitalize on prior investments. Moreover, translating data across diverse biological contexts—such as from non-invasive blood samples to tissue samples—holds significant promise for clinical practice. In toxicology, it facilitates cross-assay extrapolation, notably the translation of in vitro findings to in vivo observations, known as in vitro-to-in vivo extrapolation (IVIVE). For drug safety, it bridges the gap between preclinical animal investigations and clinical outcomes. Recent advancements in AI, particularly in GenAI, have offered novel ways to augment translational research, exemplified by techniques like Generative Adversarial Networks (GANs) and diffusion models.
Approaches: We are investigating GenAI architectures including both GANs and diffusion models in the TranslAI initiative. A pilot study using a GAN method led to a model—“TransTox”—which facilitates bidirectional translation of transcriptomic profiles between the liver and kidney under drug treatment. TransTox has demonstrated robust performance validated across independent datasets and laboratories in the following ways: 1) the concordance between real experimental data and synthetic data generated by TransTox was demonstrated in characterizing toxicity mechanisms compared to real experimental settings and 2) TransTox proved valuable in gene expression predictive models where synthetic data could be used to develop gene expression predictive models or serve as "digital twins" for diagnostic applications. The TransTox approach holds potential for multi-organ toxicity assessment with AI and advancing the field of precision toxicology.
Potential Impact: There are several challenges in regulatory science, including limited available data on a topic, translation of findings across experimental models, and merging findings from various platforms to power analyses. This initiative will provide preliminary data supporting the potential use of advanced GenAI models for experimental data translation, a potential method to extend the data diversity, and a possible approach to translating data between experimental platforms.
References
| Year | Title | Authors | Full Citation |
|---|---|---|---|
| 2024 | Bridging Organ Transcriptomics for Advancing Multiple Organ Toxicity Assessment with a Generative AI Approach. | Li T., Chen X., and Tong W. | Bridging Organ Transcriptomics for Advancing Multiple Organ Toxicity Assessment with a Generative AI Approach. Li T., Chen X., and Tong W. npj Digit. Med. 2024, 7, 310. https://doi.org/10.1038/s41746-024-01317-z |
| 2023 | TransOrGAN: An Artificial Intelligence Mapping of Rat Transcriptomic Profiles between Organs, Ages, and Sexes. | Li T., Roberts R., Liu Z., and Tong W. | TransOrGAN: An Artificial Intelligence Mapping of Rat Transcriptomic Profiles between Organs, Ages, and Sexes. Li T., Roberts R., Liu Z., and Tong W. Chemical Resesarch in Toxicology. 2023, 36(6):916-925. doi: 10.1021/acs.chemrestox.3c00037. |