Where AI meets biology

NEW YORK—In July, VantAI and TARA Biosystems announced the launch of their new biology-driven, artificial intelligence (AI)-enabled collaboration for accelerated cardiac drug development. This dynamic partnership will leverage TARA’s state-of-the-art, in-vitro human biology models and VantAI’s leading computational drug discovery capabilities to identify and develop new therapies to fight cardiac disease. VantAI provides machine learning-enabled gene-disease mapping techniques, which can be combined with the high-fidelity phenotypic data generated from TARA’s in-vitro human cardiac disease models, allowing the team to identify novel drug targets linked to cardiac disease.

Once this is done, VantAI’s in-silico platform, which can design and model both traditional small molecules and more exotic chemistry, will have the ability to produce novel molecules that effectively modulate these targets, many of which have been previously considered “undruggable.” With this information, VantAI’s platform will prioritize modality strategies and use generative approaches to design precise chemistry that is effective while also displaying favorable pharmaceutical properties.

The functional response of these novel molecules can then be evaluated in TARA’s induced pluripotent stem cell-derived human cardiac tissue models, which include a repertoire of healthy, gene-edited, patient-derived and drug-induced phenotypes of human disease. These models represent a versatile platform for capturing physiologic endpoints of human cardiac function, including contractility, electrophysiology, calcium signaling and structure, as well as genomic, proteomic and metabolic profiles. The comprehensive experimental results will be fed back to VantAI’s in-silico platform to both power further refinement of lead candidates and expand the VantAI-TARA joint data graph view of cardiac disease systems.

“The VantAI-TARA partnership is a timely demonstration of the transformative potential of biology-driven, AI-enabled drug discovery and development,” said Dr. Misti Ushio, CEO of TARA Biosystems. “The biological relevance of VantAI’s models make it an ideal development partner and through our closely integrated ‘in-silico, in-vitro’ approach, we look forward to being able to more rapidly deliver effective cardiac therapies to patients.”

While the partnership took a while to establish, Ushio tells DDN that the two likeminded companies were an extremely compatible match, and talks moved from how TARA could help VantAI by offering data points to how TARA could use VantAI’s chemistry expertise to drive their projects.

By weaving together VantAI’s in-silico technology and TARA’s in-vitro human biology, the companies have created a high-throughput feedback loop for identifying and optimizing potential drug candidates, which has already produced hits for several drug targets and promises to generate many more. Using the combined strengths of the two companies, these molecules can be optimized to rapidly advance clinical-stage drug candidates.

Additionally, part of the collaboration involves TARA working with VantAI to generate custom datasets specifically designed to improve VantAI’s existing training data, which in turn will center around molecules chosen for their ability to maximize machine-learning model performance and applicability domain, representing a novel investment in data generation that will enable VantAI’s models to achieve previously unattainable levels of accuracy and predictive power. The use of TARA’s tissue engineering technology, wherein stem cells are taken from a blood sample and used to create muscle and tissue, minimizes the number of human subjects needed for training and testing and allows VentAI to test the efficacy of their programs.

“Working with TARA allows us to build better models than ever before, and then rapidly validate our in-silico predictions in the lab,” commented Zach Carpenter, president of VantAI. “TARA’s industry-leading technology platform and the TARA team’s deep expertise in cardiac disease biology provide the perfect complement to our computational capabilities.”

Ushio says this collaboration is just the beginning. There are already a number of new projects in the works, with plans to focus on areas that are historically hard to drug or test. This partnership means that even if some of these projects fail, failing fast makes it possible to try additional projects faster, resulting in more and more targets being identified. Smaller time frames mean less commitment, and this makes it easier to go back to the beginning with new information. Ushio hopes this bespoke approach to data collection, analysis and usage can be used for developing training sets and test predictions, whether or not they are TARA’s projects.