Measuring the Effectiveness of Gene Clustering Algorithms
Author Information
Author(s): Loganantharaj Raja, Cheepala Satish, Clifford John
Primary Institution: Bioinformatics Research Lab, University of Louisiana at Lafayette
Hypothesis
Can a new metric improve the effectiveness of clustering algorithms for DNA microarray expression data by relating it to biological features?
Conclusion
The proposed metric provides a novel approach to gauge the effectiveness of gene clustering by using characteristics such as molecular function and biological processes.
Supporting Evidence
- The proposed metric measures inter-cluster cohesiveness and intra-cluster separation based on biological features.
- Results showed that genes with similar expression profiles are more closely related to biological processes than molecular functions.
- The metric can be extended to other gene features like DNA binding sites and protein-protein interactions.
Takeaway
This study created a new way to check if groups of genes are similar based on their functions, helping scientists understand how genes work together.
Methodology
The study used hierarchical and k-means clustering algorithms with Euclidean and Pearson correlation distances to analyze gene expression data.
Limitations
The metric may not account for genes that map onto multiple functions or processes.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website