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

Neural networks for secondary structure and structural class predictions

J M Chandonia1, M Karplus

  • 1Biophysics Program, Harvard University, Cambridge, Massachusetts 02138, USA.

Protein Science : a Publication of the Protein Society
|February 1, 1995
PubMed
Summary

This study introduces two neural network algorithms that improve protein structure prediction accuracy. The combined approach enhances both tertiary structural class and secondary structure predictions, especially for novel proteins.

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

All-atom empirical potential for molecular modeling and dynamics studies of proteins.

The journal of physical chemistry. B·2014
Same author

A conformational transition in the myosin VI converter contributes to the variable step size.

Biophysical journal·2011
Same author

Calculation of free-energy differences by confinement simulations. Application to peptide conformers.

The journal of physical chemistry. B·2009
Same author

CHARMM: the biomolecular simulation program.

Journal of computational chemistry·2009
Same author

Molecular dynamics studies of NMR relaxation in proteins.

Biophysical journal·2009
Same author

Protein-Protein Interactions in DNA Recognition: H-NMR Studies of Lambda cI Repressors Genetically Altered by Site-Directed Mutagenesis.

Biophysical journal·2009

Area of Science:

  • Computational biology
  • Bioinformatics
  • Machine learning in structural biology

Background:

  • Accurate prediction of protein structure is crucial for understanding biological function.
  • Existing methods for predicting tertiary structural class and secondary structure have limitations.
  • Neural networks offer a powerful approach for complex biological predictions.

Purpose of the Study:

  • To develop and evaluate a pair of integrated neural network algorithms for protein structure prediction.
  • To improve the accuracy of predicting both the tertiary structural class and secondary structure of proteins.
  • To investigate neural network optimization techniques for biological sequence analysis.

Main Methods:

  • Development of two interconnected neural network-based algorithms.

Related Experiment Videos

  • Application of algorithms to predict tertiary structural class and secondary structure.
  • Examination of neural network optimization aspects, including overtraining and rolling average filters.
  • Utilizing "jackknife" cross-validation for unbiased accuracy assessment.

    • Main Results:

    • Significant improvement in tertiary structural class prediction accuracy for nonhomologous proteins (62.3% to 73.9%).
    • Slight improvement in secondary structure prediction accuracy (62.26% to 62.64%).
    • Demonstration of synergistic accuracy gains between the two algorithms.
    • Highlighting the importance of jackknife cross-validation for reliable secondary structure prediction.

    Conclusions:

    • The integrated neural network approach enhances protein structure prediction capabilities.
    • The method shows particular promise for predicting the structure of novel proteins.
    • Further optimization of neural networks and validation strategies are key for advancing protein structure prediction.