Microarray technology is a big step in bioinformatics. Hidden information within the large amounts of data provides scientists with molecular functions or essential biological meanings to study and analyze. However, these data often contain a certain portion of entities that are missing. Several methods to estimate these missing values are developed, but most of them are with disadvantages. In this paper, we propose a novel approach to deal with these missing values based on a practical similarity measurement between gene pairs. Our approach takes gene expression values and gene ontology (GO) information for genes into consideration. We implement our approach on a real microarray dataset and compare its imputation accuracy with other methods. Experimental results show that our approach can estimate missing values in microarray data effectively.
Proceeding of IEEE International Conference on Bioinformatics & Biomedicine Workshops (BIBM 2010), pp.535-540