A Novel Approach for Handling Missing Values in Microarray Data

Abstract

The gene expression microarray is a popular technique to discover significant marker genes for different experimental design. However, missing value may occur during experimental operation or image analysis phase. Effective missing value estimation methods have been proposed to solve the problem. But, most imputation algorithms only consider the expression data in selection process. In this paper, we proposed a novel method, namely Protein and Gene Annotation K Nearest Neighbors (PGAKNN), to impute missing value of microarray gene expression data by using external biological information, like Gene Ontology Semantic Similarity and Ontology-based Protein Similarity between two genes. The experimental results show that PGAKNN provides a higher accuracy of missing value estimation on the two real yeast cDNA microarray datasets.