Folding Membrane Proteins by Deep Transfer Learning

Sheng Wang, Zhen Li, Yizhou Yu, Jinbo Xu*

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

50 Scopus citations


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-MPs and then predicts 3D 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, and generates 3D models with root-mean-square deviation (RMSD) less than 4 and 5 Å for 57 and 108 MPs, respectively. A rigorous blind test in the continuous automated model evaluation project shows that our method predicted high-resolution 3D models for two recent test MPs of 210 residues with RMSD ∼2 Å. We estimated that our method could predict correct folds for 1,345–1,871 reviewed human multi-pass MPs including a few hundred new folds, which shall facilitate the discovery of drugs targeting at MPs. A deep transfer learning method is presented to predict membrane protein contact map by learning sequence-structure relationships from non-membrane proteins, which overcomes the challenge that there are not many solved membrane protein structures for deep learning model training. The predicted contacts are pretty accurate and can help predict correct folds and accurate 3D models for ∼40% and ∼20% of 510 non-redundant membrane proteins, respectively.

Original languageEnglish (US)
Pages (from-to)202-211.e3
JournalCell systems
Issue number3
StatePublished - Sep 27 2017

Bibliographical note

Funding Information:
This work is supported by NIH grant R01GM089753 to J.X. and National Science Foundation grant DBI-1564955 to J.X. The authors are also grateful to the support of Nvidia and the computational resources provided by XSEDE . An early version of this paper was submitted to and peer reviewed at the 2017 Annual International Conference on Research in Computational Molecular Biology (RECOMB). The manuscript was revised and then independently further reviewed at Cell Systems.

Publisher Copyright:
© 2017 Elsevier Inc.


  • co-evolution analysis
  • deep learning
  • deep transfer learning
  • homology modeling
  • membrane protein contact prediction
  • membrane protein folding
  • multiple sequence alignment

ASJC Scopus subject areas

  • Pathology and Forensic Medicine
  • Cell Biology
  • Histology


Dive into the research topics of 'Folding Membrane Proteins by Deep Transfer Learning'. Together they form a unique fingerprint.

Cite this