Prediction of cis/trans isomerization in proteins using PSI-BLAST profiles and secondary structure information
2006

Predicting Proline Isomerization in Proteins

Sample size: 2424 publication Evidence: moderate

Author Information

Author(s): Song Jiangning, Burrage Kevin, Yuan Zheng, Huber Thomas

Primary Institution: The University of Queensland

Hypothesis

Can we accurately predict proline cis/trans isomerization in proteins using support vector machines and sequence information?

Conclusion

A new method using support vector machines significantly improves the prediction of proline cis/trans isomerization in proteins.

Supporting Evidence

  • Using PSI-BLAST profiles improved prediction accuracy from 62.8% to 69.8%.
  • The combination of multiple sequence alignment and secondary structure information yielded a prediction accuracy of 71.5%.
  • Support vector machines outperformed other machine learning methods in predicting proline isomerization.

Takeaway

This study helps us understand how proteins fold by predicting how a specific part of them, called proline, can twist in two different ways.

Methodology

The study used support vector machines with various sequence encoding schemes and evaluated performance through 5-fold cross-validation.

Potential Biases

Potential bias due to the dataset being limited to non-homologous proteins and the reliance on specific sequence features.

Limitations

The dataset may not represent all protein types, and the imbalance between cis and trans samples could affect prediction accuracy.

Statistical Information

Statistical Significance

p<0.05

Digital Object Identifier (DOI)

10.1186/1471-2105-7-124

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication