Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

Template-based prediction of protein structure with deep learning.

Haicang Zhang1, Yufeng Shen2,3,4,5

  • 1Department of Systems Biology, Columbia University, New York, NY, USA. hz2529@cumc.columbia.edu.

BMC Genomics
|December 29, 2020
PubMed
Summary
This summary is machine-generated.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Immunogenomic landscape of the T cell repertoire after human lung transplantation and its clinical relevance.

American journal of transplantation : official journal of the American Society of Transplantation and the American Society of Transplant Surgeons·2026
Same author

Modeling patient variants of <i>Cnot1</i> and <i>Cdc42bpb</i> results in distinct forms of congenital diaphragmatic hernia in mice.

bioRxiv : the preprint server for biology·2026
Same author

Molecular dynamics simulations of intrinsically disordered protein regions enable biophysical interpretation of variant-effect predictors.

HGG advances·2026
Same author

Machine learning to infer neurocognitive testing scores among adolescents and young adults with congenital heart disease.

Communications medicine·2026
Same author

Protein language models trained on biophysical dynamics inform mutation effects.

Proceedings of the National Academy of Sciences of the United States of America·2026
Same author

MYRF controls mesothelium specification, signaling, and plasticity in lung development.

Developmental cell·2026

ThreaderAI, a novel deep learning method, enhances protein tertiary structure prediction by improving template-query alignment. It outperforms existing methods, particularly for proteins with distant homologs, advancing biological understanding.

Area of Science:

  • Computational biology
  • Structural bioinformatics
  • Artificial intelligence in protein science

Background:

  • Accurate protein structure prediction is crucial for understanding protein function.
  • Template-based modeling (homology modeling, protein threading) is widely used but faces challenges in template selection and alignment, especially for proteins with distant homologs.
  • Developing advanced computational methods is essential to overcome these limitations.

Purpose of the Study:

  • To introduce ThreaderAI, a new template-based modeling method for improved protein tertiary structure prediction.
  • To enhance the accuracy of template-query alignment and protein threading, particularly for proteins lacking close structural homologs.
  • To leverage deep learning techniques for more precise protein structure modeling.

Main Methods:

Keywords:
Deep learningDeep residual neural networkProtein structure predictionProtein threading

Related Experiment Videos

  • ThreaderAI formulates template-query alignment as a computer vision pixel classification problem.
  • It utilizes a deep residual neural network integrating sequence profiles, predicted structural features, and contact information.
  • A dynamic programming algorithm is applied to a predicted residue-residue alignment probability matrix to build the final alignment.

Main Results:

  • ThreaderAI demonstrates superior performance in template-query alignment and protein threading compared to established methods like HHpred, CNFpred, and CEthreader.
  • Significant improvements in alignment accuracy (TM-score) were observed, especially for proteins with distant homologs.
  • On SCOPe data (fold level similarity), ThreaderAI showed 56%, 13%, and 11% higher alignment accuracy than HHpred, CNFpred, and CEthreader, respectively.
  • On CASP13 TBM-hard data, ThreaderAI outperformed HHpred, CNFpred, and CEthreader by 16%, 9%, and 8% in TM-score, respectively.

Conclusions:

  • Deep learning, as implemented in ThreaderAI, substantially enhances the accuracy of template-based protein structure prediction.
  • ThreaderAI is particularly effective for predicting structures of proteins with only distant homologs.
  • The method represents a significant advancement in computational protein structure modeling.