Noetik Secures $40 Million Series A to Accelerate AI-Driven Precision Cancer Therapies
Noetik, an emerging biotech firm specializing in the application of artificial intelligence (AI) to oncology, has announced the successful closure of a $40 million Series A financing round. The funding will support the company's efforts to advance its precision cancer therapies, leveraging its proprietary machine learning platforms and extensive spatial omics datasets.
Strategic Investors and Expanded Board
The financing round was led by Polaris Partners, with managing partner Amy Schulman set to join Noetik's board of directors. New investors, including Khosla Ventures, Wittington Ventures, and Breakout Ventures, also participated, alongside existing backers such as DCVC, Zetta Venture Partners, and Catalio Capital Management. The round further attracted investment from specialized AI funds, including ApSTAT Technologies, Linearis Labs, and Ventures Fund, with notable support from AI expert Yoshua Bengio and metabolomics authority David Wishart.
Expanding Capabilities and Platform Development
Noetik plans to utilize the Series A funds to expand its spatial omics-based atlas of human cancer biology, which is among the largest of its kind globally. This atlas is a crucial component of Noetik's strategy, providing comprehensive insights into tumor biology that inform the company's therapeutic development. Additionally, the financing will allow Noetik to scale its high-throughput in vivo CRISPR Perturb-Map platform and further train its multi-modal cancer foundation models. These models, trained using the company's unique datasets, are designed to uncover new therapeutic targets and biomarkers, significantly improving the probability of clinical success.
CEO and Co-Founder Ron Alfa, M.D., Ph.D., emphasized the importance of this funding round in advancing Noetik’s mission:
"This significant financing will enable us to accelerate our progress toward turning biological insights into a portfolio of therapeutic candidates. We are thrilled to have the support of incredible investors who share our vision of combining deep patient data and artificial intelligence to build the future of cancer therapeutics."
As part of its growth strategy, Noetik is actively seeking to establish partnerships with leading academic institutions, healthcare providers, and pharmaceutical companies. To lead these efforts, the company has appointed Shafique Virani, M.D., Ph.D., as Chief Business Officer.
Understanding Noetik's OCTO Technology
OCTO, Noetik's Oncology Counterfactual Therapeutics Oracle, represents an advance in AI-driven cancer research, employing a novel multimodal, transformer-based architecture that processes spatially aligned data from diverse modalities, such as multiplex protein staining, spatial gene expression, DNA sequencing, and histopathology images (e.g., H&E stained sections). To achieve this, OCTO integrates each data type into a single unified representation by encoding it as a sequence of biologically meaningful tokens—such as individual protein channels from fluorescence images or gene expression levels at specific tumor regions.
The model is trained using a unique form of structured multimodal masking, where nearly all data tokens are masked, compelling OCTO to infer missing information by combining insights across data modalities. This approach allows OCTO to simulate complex biological phenomena, such as predicting protein co-expression from gene expression patterns or reconstructing high-resolution protein images from sparse data.
Crucially, OCTO's training incorporates counterfactual simulations, enabling it to learn how specific molecular changes—like the knockout of a gene—impact cellular behavior and protein interactions in a spatially resolved context. For example, by manipulating gene expression patterns and measuring resulting changes in predicted protein levels, OCTO can simulate the effects of different therapeutic strategies on patient-specific tumors.
Additionally, OCTO can generalize these findings to emergent properties, such as inferring tumor and immune cell identities solely from nuclear morphology, an ability reminiscent of zero-shot learning in language models. By scaling up this multimodal training with over 20 billion tokens across hundreds of GPUs, OCTO achieves a deep, context-aware understanding of cancer biology that extends beyond simple data correlation, potentially revolutionizing how we identify and validate new drug targets through in silico experimentation and real-time hypothesis testing.
Topics: Startups & Deals
