Nature paper reports graph model predicting non-peptidic macrocycle permeability

The wire

1. 01Researchers describe MEGPNM, a multilayer edge-aware graph attention network with Jumping Knowledge, trained on PAMPA data from the Non-peptidic Macrocycle Membrane Permeability Database to predict membrane permeability.
2. 02Macrocycles are pursued for hard-to-drug targets with oral exposure potential, but conformational flexibility and chameleon behavior make permeability hard to predict, so class-specific models matter for early screening.
3. 03Attention analyses flag recurring structural motifs tied to permeability; the work is a computational benchmark on one assay dataset, not experimental validation of any candidate.

From the source

MEGPNM as a multiscale edge-aware GAT network with hybrid pooling predicting permeability of non-peptidic macrocycles

Non-peptidic macrocycles are attractive therapeutic candidates for targets that are difficult to drug, while still offering a path toward oral exposure. Predicting their membrane permeability, however, remains difficult because these molecules are conformationally flexible and can display molecular chameleon behavior. These properties make permeability prediction particularly challenging, so macro