by Katyanna Quach — theregister.com — Startup Absci will turn to its generative AI algorithms to design synthetic protein

Pharma giant AstraZeneca has signed a $247 million deal with Ai drug creation company Absci to develop an antibody designed to fight cancer. The collaboration was announced on Monday and will see Absci deliver an antibody for a “specified oncology target”. AstraZeneca will cover its R&D efforts, while Absci will make a cut from royalty fees from any future sales if the drug makes it to market. “Absci’s newest partnership with AstraZeneca is further validation of our first-of-its-kind zero-shot generative AI model designed to create new and improved antibody therapeutics, including for previously untreatable diseases,” Absci CEO Sean McClain told The Register in a statement. “We’re proud to work closely with AstraZeneca to leverage our AI to bring novel treatments to oncology patients,” he added.

Can AI transformer models help design drugs and treat incurable diseases?

Chan Zuckerberg org to spin up 1,000+ H100 GPU cluster for AI medical research Former infosec COO pleads guilty to attacking hospitals to drum up business An Absci representative declined to detail which cancer the antibody will targeting, and when it expected to deliver a candidate to AstraZeneca. Absci’s drug design platform uses computational techniques, including generative AI models, to create millions of potential synthetic antibodies. The most promising designs are grown from genetically engineered E coli bacteria cells. The synthetic antibodies designed by AI can be made by altering the DNA of the E coli cells, forcing them to create the desired protein structure. The resulting creations’ abilities to effectively bind to targets can then be tested in lab experiments, and the data used to further improve the model’s future outputs. Joshua Meier, Absci’s chief AI officer, previously told The Register that there are more possible antibody variants that can be simulated than there are atoms in the universe. To find the most promising candidates, the company trains AI algorithms to predict various traits that suggest they might be more effective at things like marking cancerous cells for the immune system to attack, or preventing tumors from growing.

Absci is looking for antibodies that won’t be rejected by a patient’s own immune system, and which can be easily manufactured at scale. Commercializing new drugs is a difficult, lengthy process that requires extensive assessment by regulators, including clinical trials, before official approval for medical use is granted. Big Pharma companies like AstraZeneca believe AI can help experts develop and screen drugs at a faster rate than traditional methods.

Last month, the pharmaceutical giant launched Evinova, a separate subsidiary business using AI and machine learning to speed up clinical trials. AstraZeneca believes the technology can be used to design experiments and predict their success rates based on historical data. Evinova will also build software and hire clinicians to support telehealth appointments to monitor patients during trials. “The future of medicine development can be accelerated with digital solutions,” Evinova CEO Pascal Soriot said in a statement. “We believe Evinova’s combination of scientific expertise and track record in developing AI-enabled digital technologies at scale, provides a real opportunity to fundamentally improve patient care, drive healthcare transformation and reduce carbon emissions.” The Register has asked AstraZeneca for comment. ®