Finding motif pairs in the interactions between heterogeneous proteins via bootstrapping and boosting
2009

Finding Motif Pairs in Protein Interactions

Sample size: 1712 publication Evidence: moderate

Author Information

Author(s): Kim Jisu, Huang De-Shuang, Han Kyungsook

Primary Institution: Inha University

Hypothesis

Can bootstrapping and boosting algorithms effectively predict interacting motif pairs in protein-protein interactions?

Conclusion

Bootstrapping is effective for generating a balanced negative data set, and the boosting algorithm can efficiently predict interacting motif pairs from protein interaction data.

Supporting Evidence

  • The boosting algorithm showed 84.4% sensitivity and 75.9% specificity with balanced data sets.
  • Motif pairs predicted were statistically significant.
  • Bootstrapping effectively controlled the size and distribution of negative data sets.

Takeaway

Scientists created a new method to help computers find patterns in proteins that interact with each other, making it easier to understand how they work together.

Methodology

Developed a bootstrapping algorithm for generating negative data sets and a boosting algorithm for finding interacting motif pairs.

Potential Biases

The prediction model may be biased if the training data sets are unbalanced.

Limitations

The method may not work well when sequence data is insufficient to predict motif pairs.

Participant Demographics

Interactions between 1,029 human proteins and 603 virus proteins.

Statistical Information

P-Value

3.13e-3

Statistical Significance

p<0.05

Digital Object Identifier (DOI)

10.1186/1471-2105-10-S1-S57

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication