Exploring Protein Functional Relationships Using Genomic Information and Data Mining Techniques
- First Online:
- Cite this paper as:
- Yang J.Y., Yang M.Q., Ersoy O.K. (2003) Exploring Protein Functional Relationships Using Genomic Information and Data Mining Techniques. In: Kaynak O., Alpaydin E., Oja E., Xu L. (eds) Artificial Neural Networks and Neural Information Processing — ICANN/ICONIP 2003. Lecture Notes in Computer Science, vol 2714. Springer, Berlin, Heidelberg
Anapproach that uses both supervised and unsupervised learning methods for exploring protein functional relationships is reported; we refer to this as Maximum Contrast (MC) tree. The tree is constructed by performing a hierarchical decomposition of the feature space; this step is performed regardless of complex nature of protein functions, i.e. it performs this decomposition even without knowledge of the protein functional class labels. In order to test our algorithm, we have constructed a library of Protein Phylogenetic Profiles for the proteins in the yeast Saccharomyces Cerevisiae with 60 species. Results showed our algorithm compares favorably to other classification algorithms such as the decision tree algorithms C4.5, C5, and to support vector machines.
Unable to display preview. Download preview PDF.