Selection of informative clusters from hierarchical cluster tree with gene classes

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Hierarchical clustering is a common method for gene expression data analysis.
• Traditional methods involve cutting the cluster tree or analyzing sorted gene lists, leading to information loss.
• The proposed method utilizes gene classifications for more effective cluster selection.

Purpose of the Study:

• To present a simple method for identifying clusters with significant gene class enrichment from a cluster tree.
• To demonstrate the method's ability to find informative clusters at various levels of the tree.
• To compare cluster trees generated by different clustering methods.

Main Methods:

• A novel method for searching clusters based on gene class enrichment within a cluster tree.
• Application of the method to a yeast gene expression dataset and two database classifications.
• Visualization of results overlaid on the cluster tree.

Main Results:

• The method successfully identified clusters with strong enrichment of functional gene classes.
• Discovered gene groups similar to and additional to those found in the original analysis.
• Demonstrated that informative clusters can be found at multiple tree levels, surpassing simple tree cutting.

Conclusions:

• The developed method facilitates exploratory analysis of large datasets with available categorical data.
• It offers a more comprehensive approach to cluster identification compared to traditional methods.