📢 Mosaic Biosciences is thrilled to join the AIntibody initiative alongside industry pioneers! This innovative competition will put AI-driven antibody discovery to the test, evaluating its ability to design antibodies against specific epitopes without traditional experimental methods. At Mosaic, we have implemented industry-leading developability assays to complement our antibody discovery and optimization capabilities. We are eager to contribute these assays to this exciting and important initiative. Special thanks to our collaborators, including Specifica, an IQVIA business, Bio-Techne, Azenta Life Sciences, Carterra, and Sapidyne Instruments Inc. for making this initiative possible. 🔗 Discover more about this exciting challenge and its potential to advance antibody discovery technology here: https://lnkd.in/g2p3WWUP #AntibodyDiscovery #AIntibodyChallenge #BiotechInnovation #AI
Founder @ Specifica | Antibody Libraries / Antibody Discovery and Optimization | MD PhD
Is AI a disruptive technology in Antibody Discovery? Or is it just addressing already solved problems? Can it even do what its proponents claim? It is now relatively straightforward to generate antibodies against essentially any target using in vivo or (preferably) in vitro approaches. Recent publications have described methods to generate antibodies against specific epitopes. We have generated highly developable antibodies directly from our Generation 3 libraries with affinities as high as 11pM without affinity maturation. Improvement of affinities and developability using experimental methods is relatively straightforward. So, what does AI/ML promise that we can’t already do? I believe it will eventually be able to design antibodies binding specific epitopes, but find it difficult to believe we will ever be able to avoid experimental confirmation, and in that case, how much faster will it be than a well-equipped lab, or mouse? Within the context of a 5-10 year drug development cycle, how much difference will it make? As the number of AI/ML publications and talks in antibody discovery proliferate, many of which I don’t fully understand, I find myself oscillating between thinking, “time to retire, AI/ML is going to put all us experimentalists out of business”, and “this is yet another hype cycle”. So, we decided to put it to the test. In the first of what we anticipate will be an annual competition akin to CASP (which demonstrated the value of Alphafold), today we announce the launch of AIntibody, an experimentally validated in silico antibody discovery design challenge. These first challenges are based on two NGS datasets Specifica, an IQVIA business generated against the RBD of SARS-CoV-2, which we anticipate should make these first challenges relatively straightforward given the amount of data on RBD. Read about the challenges in the paper. A lot of enthusiasm for this initiative with support from Azenta Life Sciences, who will make the antibodies; Carterra and Sapidyne Instruments Inc., who will measure antibody affinities; Mosaic Biosciences will assess developability; and Bio-Techne who will provide the target; not to mention our co-authors M Frank Erasmus, Laura Spector, Fortunato Ferrara, Roberto di Niro, Tom Pohl, Katheryn P., Wei Wang, Peter Tessier, Crystal Richardson, Ph.D., Laure Turner, Sumit Kumar, Ph.D, Daniel Bedinger, Pietro Sormanni, Monica Fernández-Quintero, Andrew Ward, Johannes Loeffler, Olivia Swanson, Charlotte Deane, Matthew Raybould, Andreas Evers, Carolin Sellmann, Sharrol Bachas, Jeffrey Ruffolo, Horacio Nastri, Karthik Ramesh, Jesper Sørensen, Rebecca Croasdale-Wood, Oliver Hijano Cubelos, Camila Leal, Melody Shahsavarian and others below. See the article here: https://lnkd.in/gfMz7JCh