The Prediction of Membrane Protein Types by K-Substring Source of Diversity and Weighted-KNN Algorithm

姜彬,王正华,王勇献,贺细平
DOI: https://doi.org/10.3969/j.issn.1007-7146.2009.01.025
2009-01-01
Abstract:Membrane proteins is the main manifestation of biological membrane's function.It plays a crucial role in cells and makes the material basis for cells to implement various functions.The prediction for the type of membrane protein is a crucial fundamental research in the field of the structure and function of membrane protein and will also provide guidance for the related research in biology.In order to predict the type of membrane protein,this paper uses the method of k-substring source of diversity to extract the feature of membrane protein.Meanwhile we construct a new type of membrane protein classification model that combines the approach of the smallest increment of diversity with the weighted-KNN algorithm.Under three typical methods(Self-consistency,Jackknife and Independent dataset),the accuracy rate of our prediction is respectively 99.95 %,86.16 % and 98.36 %.The experimental results demonstrate the usefulness of above method to extract the characteristic information and predict the type of membrane protein.
What problem does this paper attempt to address?