Photo by National Cancer Institute on Unsplash
Bio Information is Energy
COGNANO is a venture that aims at computer-aided drug discovery. The large VHH data obtained from alpacas will lead us to a new drug discovery platform. Our goal is to optimize drug design/development. For this purpose, we have brought together a large number of bio-experts and IT engineers. We are using machines to create "antibodies that can withstand all new coronary mutations" that did not exist in the world. With this opportunity, we will continue to search for and publish "unknown antigen molecules (UNMET NEEDS)" that have never been seen before by mankind.
12Years
Research on VHH Antibodies
50Types
Antigens
300MReads
Antibody Genes
30MClones
Labeled Antibody Sequences
Photo by Who’s Denilo ? on Unsplash
Pitch
COGNANO merges bio and IT by using the natural immune system. Our AI analyzes 400 million data points on antibody genes, understanding protein-antibody relations via natural language processing(LLM). Accepted at a top AI Conference NeurIPS 2023, collaborating with Google, we now explore using this technology in drug discovery based on our super big data. The strength of antibodies' ability to detect genetic and structural changes lies in their adaptability to medicine.
VHH Antibody
For DNA organisms including humans, the genome should not change throughout our lives, but only the antibody genes change almost infinitely (hypermutation), and the acquired antibodies fight viruses, bacteria, and cancer. In this meaning, antibodies are amazing nature's masterpiece. Evolutionarily, it originated in cartilaginous fish such as the eight-eyed eel, then evolved into a more complex fo...
Projects
Blog
Akihiro Imura
Hirofumi Tsuruta
Hiroyuki YamazakiSynergy between immune response and machine learning
Vertebrates have the ability to make complex (full-bodied) antibodies composed of light and heavy chains, but only camelids have single-domain antibodies in addition to full-bodied antibodies. Single domain antibodies can be easily made into a gene library from lymphocyte genes, and antibodies created by immunized camelids will be big data.
Biopanning-labeled data using intermolecular interactions (antigen-antibody reactions) are integrated by COGNANO's unique algorithm, making it possible to predict the binding characteristics (binding affinity, epitope) to antigen molecules.COGNANO's original platform that can calculate antibody characteristics by machine learning huge amounts of real antibody data and provide them as seeds at high speed.
Even if the naive library, which is not immunized, is analyzed, it cannot be used as teaching data for machine learning because the number of types of binding antibodies is too small. Our Cognano VHH platform make it possible to discover target antibodies 10K times effectively compared with conventional mouse or human antibody platforms.