Folding membrane proteins by deep transfer learning
This work addresses the problem of limited solved membrane protein structures for drug discovery, representing a strong specific gain in computational biology.
The researchers tackled the challenge of predicting membrane protein structures by developing a deep transfer learning method that uses non-membrane proteins to predict contacts and generate 3D models, achieving a contact prediction accuracy improvement of at least 0.18, correct folds for 218 proteins, and high-resolution models with RMSD close to 2 Angstrom in blind tests.
Computational elucidation of membrane protein (MP) structures is challenging partially due to lack of sufficient solved structures for homology modeling. Here we describe a high-throughput deep transfer learning method that first predicts MP contacts by learning from non-membrane proteins (non-MPs) and then predicting three-dimensional structure models using the predicted contacts as distance restraints. Tested on 510 non-redundant MPs, our method has contact prediction accuracy at least 0.18 better than existing methods, predicts correct folds for 218 MPs (TMscore at least 0.6), and generates three-dimensional models with RMSD less than 4 Angstrom and 5 Angstrom for 57 and 108 MPs, respectively. A rigorous blind test in the continuous automated model evaluation (CAMEO) project shows that our method predicted high-resolution three-dimensional models for two recent test MPs of 210 residues with RMSD close to 2 Angstrom. We estimated that our method could predict correct folds for between 1,345 and 1,871 reviewed human multi-pass MPs including a few hundred new folds, which shall facilitate the discovery of drugs targeting at membrane proteins.