Proteinea is a TechBio employing proprietary AI models and computational tools to engineer next-gen antibodies with improved efficacy, safety and convenience
Our pipelines utilize an ensemble of a variety of diverse learning objectives to overcome computational bias. For hits generation, we simultaneously operate on sequence and structure spanning our reinforcement learning-editors, deep learning-based fitness landscape reconstructors, and rational design proprietary frameworks.
.jpeg)
We leverage the consensus of state-of-the-art physics-based tools, AI-based predictors, and molecular dynamics pipelines to evaluate the efficacy, yield, safety, and pharmacological properties of our generated hits.
.jpeg)
We promote knowledge-guided optimization by iterative feeding back from both computational and experimental validation pipelines. We utilize our lab facilities specialized in the production and functional testing of therapeutic proteins with smart experimental throughput.
Ankh is the first general-purpose protein transformer trained on Google's TPU v4 that exceeds the performance of the current world best transformers in a representative poll of structure and function benchmarks with less than 10% of the parameters and 30% of the embedding dimension.
Explore the featured white-papers for a deeper look at some of our innovations
Fc-SubQer offers novel compositions and methods involving the modification of the constant region of antibodies. These compositions have been designed to ensure that they do not have any detrimental effects on important properties of antibodies, including affinity, stability, and biological functions such as effector function and half-life
.png)
