Imputing Missing Data for Gene Expression Arrays

Abstract

Here we describe three different methods for imputation.The first is based on a reduced rank SVD of the expression matrix, the second is based on K-nearest neighbor averaging, and the third is based on repeated regressions. We demonstrate the techniques on the human tumor data and a subset of the yeast data. 1 Imputation using the SVD The singular value decomposition offers an interesting and stable method for imputation of missing values in gene expression arrays.The basic paradigm is • Learn a set of basis functions or eigen-genes from the complete data. • Impute the missing cells for a gene by regressing its non-missing entries on the eigen-genes, and use the regression function to predict the expression values at the missing locations.