ABSTRACT: Prediction of protein function is important in studying biological processes, and can be challenging when proteins lack sequence similarity to well-characterized proteins. We have developed a web-based software that uses the SVM (Support Vector Machine) machine learning method for classification of a protein, irrespective of alignment-based similarity, into functional family using its primary sequence. This software, SVMProt, may be used as a protein function prediction tool that complements sequence alignment methods. SVMProt can be accessed at http://jing.cz3.nus.edu.sg/cgi-bin/svmprot.cgi. The theory and construction of SVMProt will be discussed, along with its performance with selected datasets.
ABOUT THE OLS BIOINFORMATICS FOR BIOLOGISTS SEMINAR SERIES:
The "virtuous cycle" of bioinformatics is revolutionizing the
biomedical sciences. Computational biologists produce analytical
and predictive bioinformatics tools that are increasingly used
to guide wet-lab research. In turn, the wet-lab biologist has
valuable experience, knowledge and data that can be used to
refine those bioinformatics tools.
The OLS Bioinformatics for Biologists seminar series will bring
together wet-lab and computational biologists from NUS, in an
informal setting, to promote collaborations and exchange of ideas.
Selected NUS computational biologists whose research has practical
application to wet-lab biology will present their work. Active
exchange of information during the seminar is encouraged.