Improved residue contact prediction using support vector machines and a large feature set
2007

Improved Protein Contact Prediction with SVMs

Sample size: 48 publication Evidence: moderate

Author Information

Author(s): Cheng Jianlin, Baldi Pierre

Primary Institution: University of Central Florida

Hypothesis

Can support vector machines improve the accuracy of protein residue contact predictions?

Conclusion

The SVMcon predictor shows a 4% improvement in accuracy over the previous best method and ranks among the top predictors in the CASP7 experiment.

Supporting Evidence

  • SVMcon outperformed the latest version of the CMAPpro contact map predictor by 4%.
  • SVMcon was ranked as one of the top predictors in the CASP7 experiment.
  • The method integrates various features including profiles and secondary structure for improved predictions.

Takeaway

Scientists created a new tool to better predict how parts of proteins interact, which helps in understanding their structure.

Methodology

The study used support vector machines trained on a dataset of protein sequences to predict medium- and long-range contacts.

Limitations

The accuracy of predictions varies significantly with different proteins and their structure classes.

Digital Object Identifier (DOI)

10.1186/1471-2105-8-113

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication