Ai startups

atomwise origins is chess Computation biology design of a pharma molecule. Machine learning and big data sets and biology kept creating molecules in simulation that worked. For a drug to work it needs to stick to a protein and arrest the function of the protein. Other proteins need to continue to function properly with the molecule.  The problem becomes a prediction problem. Does the molecule stick, yes or no? test the molecule will bind to the protein. the drug molecule blocks up the protein. Proteins are 3d matrix: oxygen, carbon, hydrogen extra. The 3d grid is feed to the deep learning network cnn. institutions with specific proteins can have a list of possible molecules that will bind to it. atomwise does the computations and sends the results to the partner institution which physically tests the molecule list and sends atomwise back the results.