Penalized likelihood for sparse contingency tables with an application to full-length cDNA libraries
2007

New Methods for Analyzing Sparse Contingency Tables in Biology

Sample size: 250 publication Evidence: moderate

Author Information

Author(s): Corinne Dahinden, Giovanni Parmigiani, Mark C. Emerick, Peter Bühlmann

Primary Institution: ETH Zürich

Hypothesis

How can we effectively analyze interactions among categorical variables in high-dimensional biological data?

Conclusion

The study presents regularization methods that successfully detect complex interaction patterns among categorical variables in biological datasets.

Supporting Evidence

  • The proposed methods were tested on simulated data and showed improved performance in model selection.
  • The study highlights the importance of regularization techniques in analyzing high-dimensional biological data.
  • Results indicate that the level-ℓ1-regularization method outperforms traditional methods in terms of model selection accuracy.

Takeaway

The researchers created new ways to look at how different parts of genes interact with each other, which helps us understand how genes work together.

Methodology

The study developed methods for model selection and parameter estimation in log-linear models for sparse contingency tables, using ℓ1-penalization and comparing it to other procedures.

Limitations

The methods may not be applicable to all types of biological data and rely on the assumption of sparsity in the contingency tables.

Digital Object Identifier (DOI)

10.1186/1471-2105-8-476

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication