New AI could help scientists discover new antibodies to fight cancer

Nature communication published a study on 28 January 2023, citing that a group of researchers from the University of California, San Diego, has designed a unique machine learning algorithm capable of dictating new antibodies to fight cancer. The recent discovery of an antibody that can bind the famous programmed death ligand 1 (PD-L1) 17 times better than the previously approved atezolizumab is the culmination of the effectiveness of the newly designed machine.

The new AI works by analyzing half a million possible antibody sequences in its library and screening them for their affinity to a specific protein target, unlike the traditional method of designing antibodies and growing them in bacterial or yeast cells and then testing them countless times to see how they bind to the target proteins. It takes less time and resources to do the analysis. One of its uniqueness is that it can confidently report the certainty of each prediction.

Cancer has claimed millions of lives and this latest discovery will help to reduce its threat. Researchers now have a better approach to the discovery of antibodies against cancers and other important antigens such as SARS-CoV-2.

They are also programming another AI model that will be able to analyze amino acid sequences for other antibody properties important to the success of clinical trials, such as stability, solubility and selectivity.