Antibody discovery technologies have delivered life saving therapies over the past three decades, however, many challenging targets remain undrugged due to limitations of these technologies. Phage display technology presents a solution to discovering antibodies to targets that do not elicit a strong immune response in animals, are toxic, or are difficult to produce in large quantities in the correct conformation. Synthetic libraries, in which antibodies are designed in silico, are a promising approach that does not rely on animal immunizations or donors. However, most synthetic libraries in use today were constructed by fixing certain regions of the antibody while generating random sequences from position-specific amino acid frequencies for antigen-binding regions. This departure from the space of natural antibodies results in antibodies with poor biophysical characteristics such as low melting temperature, aggregation propensity, or general stickiness.Digital Proteomics has developed a computational workflow for natural antibody repertoire analysis, harnessing the throughput of next-generation sequencing of immune cells, that will be used to design a better synthetic library for therapeutic antibody discovery. Rules of natural antibody development will be incorporated into the design of a synthetic antibody library, thereby retaining the biophysical property profile of natural antibodies. Antibody genes develop through a sequential process involving site-specific genome recombination and somatic hypermutation, which will be modeled in silico. While in conventional antibodies, this process occurs at two independent loci (a heavy and light chain that must interact), the proof-of-concept will be applied to the single chain antibodies produced by camelids. Using an improved synthetic library framework, the computational design of an antibody library optimized for therapeutic discovery will contain antibodies that display superior biophysical properties and reduced sequence liabilities. The library will be an invaluable tool for rapid discovery of therapeutic antibodies to a wide range of diseases. Public Health Relevance Statement Mining large antibody libraries enables the rapid discovery of therapeutics against challenging targets, however, current libraries have had limited success in delivering antibody therapeutics due to poor biophysical characteristics such as low melting temperatures, aggregation propensity, or general stickiness. We propose the computational design of a better library that uses an in silico model of natural antibody development. Combined with computational filters to remove liabilities that preclude development as a therapeutic, our synthetic library will deliver antibodies to treat a myriad of conditions including cancer, infectious diseases, and autoimmune diseases.
