PMIpred predicts protein-membrane interactions for peptides and proteins
With increasing membrane-binding affinity, proteins can be classified as “non-binders” (stay in solution), “sensors” (only bind to curved membranes), and “binders” (bind to membranes without curvature selectivity)1. We performed MD simulations to quantify the affinity of >50,000 peptides for lipid packing defects on the surface of curved/stretched membranes and trained a neural network on these data.
You can use PMIpred to predict membrane interactions for either peptides (7-24 amino acids) or protein structures (uploading a PDB-file).
When using PMIpred, please cite: