BioMap Launches xTrimoAAVViability v1.0, an AI-Powered AAV Assembly Efficiency Prediction Model, in Collaboration with Porton Advanced
BioMap
Developer of Innovative Drug R&D Platform
Since November 2023, BioMap announced a strategic partnership with Porton Advanced. Both parties are committed to utilizing cutting-edge large-scale life science models to advance the development of algorithms related to adeno-associated virus (AAV).
After six months of close collaboration, BioMap and Porton Biologics Ltd. reached a significant milestone in July 2024 — the successful launch of an AI-driven AAV solution.
Co-creating a New Chapter in AAV Design
Based on BioMap's AIGP platform and xTrimoPGLM large model, combined with Porton Biologics Ltd.'s extensive professional experience in AAV libraries and experimental validation, a highly efficient and precise one-click AAV library design service has been developed.
This innovative achievement enables users to efficiently design and optimize AAV libraries covering more than 10 serotypes, including point mutations, short peptide insertions, and fragment recombination, to achieve higher transduction efficiency, tissue and organ specificity, and immune escape capabilities.
BioMap's xTrimoPGLM large model, which integrates a vast amount of multimodal protein data (sequence, structure, and other properties), provides a unified pre-training framework and foundational model for AAV design.
Based on this large model, a series of AI models are employed to comprehensively optimize key indicators such as assembly efficiency, tissue targeting specificity, and immune escape capability, ensuring that the generated AAV sequences exhibit higher functional performance and diversity, significantly improving the library screening success rate.
The platform demonstrates strong generalization capabilities, enabling high-precision AAV function prediction even in unseen mutation regions and serotypes.
In addition, BioMap and Porton Biologics Ltd. will also provide customized AI design services based on customers' personalized needs and specific data, fine-tuning models according to unique customer data to achieve more precise and personalized design optimization.
xTrimoAAVViability v1.0 Assembly Efficiency Model
Assist in the Construction of Library Screening and Candidate Variant Characterization
Based on BioMap's xTrimo, a life science foundational large model with hundreds of billions of parameters, and the 20 million AAV assembly efficiency data provided by Porton Biologics Ltd., BioMap has launched the industry-leading assembly efficiency prediction model —— xTrimoAAVViability v1.0. This model enables users to rapidly and accurately develop AAV vectors tailored to specific application needs, effectively reducing R&D costs and time while enhancing the market competitiveness of gene therapy products.
This R&D achievement was exhibited at the 27th Annual Meeting of the American Society of Gene and Cell Therapy (ASGCT) in 2024 and is now officially launched in the market, providing professional services to industry clients and research institutions.
In the model performance test, BioMap used the VR2 region mutation data of the AAV5 serotype for training. The results showed that the model achieved a Pearson correlation of 0.9 within the same serotype and mutation region, approaching the upper limit of biological reproducibility.
Moreover, the model still demonstrated excellent generalization ability in mutation regions not seen in the same training set AAV5 serotype or other serotypes not covered by the training set, achieving Pearson correlations of 0.71 and 0.59, respectively.
Moreover, BioMap's general protein naturalness model, xTrimoStability, achieved a Pearson correlation of 0.53 in this prediction task without any exposure to AAV data. This fully demonstrates the foundational generalization capability brought by large protein language models and provides an excellent basis for targeted fine-tuning.
xTrimoAAVViability v1.0 Assembly Efficiency Model is Now Available on BioMap AIGP Platform: https://aigp.biomap.com
The success of this collaboration heralds a significant developmental phase in the field of AAV design. By integrating AI technology with biopharmaceutical practices, BioMap and Porton Biologics Ltd. have not only greatly enhanced the efficiency and success rate of AAV screening but also provided a powerful tool and solid support for research and application in the fields of gene therapy and biomedicine.
Looking ahead, the two parties will continue to strengthen their cooperation, rapidly transforming innovative concepts into feasible candidate variants, further expanding and optimizing AAV design services, and providing the biopharmaceutical industry with more efficient and precise AAV solutions.
—The End—
Recommended Reading
