# The Biggest Challenge for Prediction of Membrane Permeability of Cyclic Peptides: Scarce Data Availability

## TL;DR

The paper discusses how limited data hinders AI predictions of cyclic peptides' membrane permeability, but a new AI model and high-throughput assay offer a solution.

## Contribution

The novel contribution is an AI model combined with a high-throughput assay to design permeable cyclic peptides with polar groups.

## Key findings

- Limited biological data hampers AI predictions of cyclic peptide membrane permeability.
- A high-throughput assay and AI model enabled classification of billions of cyclic peptides.
- De novo permeable cyclic peptides with polar groups were successfully designed.

## Abstract

AI-based multiparameter optimization for membrane permeability
of cyclic peptides is hampered by the limited availability of ground
truth biological data. This Viewpoint highlights the Featured Article
reporting an AI model, integrating a high-throughput assay classifying
billions of cyclic peptides for permeability-related characteristics.
Combining this with a generative AI, the first de novo designed permeable
cyclic peptides containing polar groups were reported.