When AI Meets Experimental Biology: Supporting Breakthrough Peptide Research

 The PEGASUS model by 1910 represents a breakthrough in AI-driven drug design, achieving something remarkable – the first reported macrocyclic peptides with more than two polar or charged fragments to demonstrate cell membrane permeability.

Why this matters:

Cyclic peptides exhibit superior drug‑like properties, including higher target‑binding affinity and excellent metabolic stability compared with linear peptides. These advantages make oral administration or intracellular targeting possible and give cyclic peptides substantial market potential. Despite these benefits, major obstacles, such as poor metabolic stability and limited membrane permeability, still need to be overcome.

Using AI for drug design can significantly accelerate the discovery process, enhance molecular design quality, and improve the accuracy of membrane‑permeability prediction. However, true experimental data are required to confirm “successful model cases.” By evaluating newly generated peptides from AI screening using MDCK‑MDR1 cell permeability assays, researchers can demonstrate that AI is capable of designing highly polar yet membrane‑permeable cyclic peptides. These experiments validate that the physicochemical features predicted by PEGASUS (e.g., logD, conformation) are consistent with real permeability outcomes, and they serve as the final benchmark to determine which AI‑designed peptides are viable for further development.

The ripple effect: Unlike traditional early‑stage permeability screens, experimental data used for AI model validation must come from reliable, standardized, and reproducible in vitro permeability systems, meet stringent QC and recovery requirements, remain within the assay’s effective detection range, and be generated efficiently.

Our DMPK team brings years of experience in in vitro ADME studies and a rich history of high‑quality experimental datasets.
We maintain a stable, efficient, and robust assay platform and can provide customized experimental designs tailored specifically to support AI model validation. As AI-driven drug design continues to advance, the need for reliable, specialised in vitro ADME assays only grows. We’re excited to support this intersection of computational and experimental science

Interested in permeability studies for your AI-designed peptides or other novel modalities? Our DMPK team brings decades of expertise in specialised assays that generate the quality data AI models need.

Read the PEGASUS publication: https://pubs.acs.org/doi/10.1021/acs.jmedchem.5c01836

Learn more about 1910