This presentation features the development of stapled helical peptide drugs, termed Helicons, and the application of computational techniques from data science and machine learning to enhance their discovery and development. John Santa Maria began by explaining the significance of alpha helices in protein secondary structures, highlighting their role in mediating cellular biology and disease processes. These helices were particularly effective in spanning membranes and reducing exposed polarity, making them suitable for drug development.
Santa Maria discussed the integration of wet and dry lab discovery platforms in the optimisation of drugs like FOG-001. This process involved helical phage display screens, computational approaches, and machine learning to predict binding sites and design peptides. He emphasised the modular nature of peptide synthesis, which allowed for the rapid synthesis of thousands of peptides and their assessment for binding and cellular permeation.
The presentation highlighted the use of machine learning models to predict peptide permeation into cells and optimise their binding to targets. These models facilitated the design, make, and test cycles for helical peptides, generating high-quality data for further development. Santa Maria provided examples of how regression models helped interpret experimental data and identify key positions for binding, leading to significant improvements in peptide potency.
He also discussed the use of generative approaches to enhance peptide properties, such as improving cellular permeation while retaining target binding. This involves computationally generating new peptide sequences and assessing their effectiveness. Santa Maria concluded by highlighting the role of data science and AI in expediting the design and optimisation of helical peptides.
