Generating confidence intervals on biological networks
2007

Generating Confidence Intervals on Biological Networks

publication Evidence: moderate

Author Information

Author(s): Thorne Thomas, Stumpf Michael PH

Primary Institution: Imperial College London

Hypothesis

Can a new network resampling Null model that incorporates biological annotations improve the statistical analysis of biological networks?

Conclusion

An improved statistical approach for the analysis of biological network data leads to qualitatively different results compared to traditional methods that ignore biological annotations.

Supporting Evidence

  • The GOcardShuffle approach results in a Null model that better describes the properties of real biological networks.
  • Incorporating GO annotations leads to a systematic shift in the statistical significance of correlations among interacting proteins.
  • The study demonstrates that conditioning on biological information can change the expected Null distribution.

Takeaway

This study created a new way to analyze biological networks that uses extra information about the proteins, which helps to get better results than older methods.

Methodology

The study introduced a new network resampling Null model called GOcardShuffle that incorporates biological annotations to assess statistical significance in protein interaction networks.

Potential Biases

There is a risk of circular arguments if in-silico annotations are used without proper validation.

Limitations

The study acknowledges that molecular network data sets are still plagued by noise and incompleteness.

Digital Object Identifier (DOI)

10.1186/1471-2105-8-467

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication