Gene selection for classifications using Multiple PCA with Sparsity

A gene selection algorithm was developed using Multiple Principal Component Analysis with Sparsity (MSPCA). The MSPCA algorithm is used to analyze normal and disease gene expression samples and to set these component loadings to zero if they are smaller than a threshold for sparse solutions. Next, genes with zero loadings across all samples (both normal and disease) are removed before extracting feature genes. Feature genes are genes that contribute differentially to variations in normal and disease samples and, thus, can be used for classification. The MSPCA is applied to three microarray datasets to select feature genes with a linear support vector machine to evaluate its performance. This method is compared with several previous gene selection results to show that this MSPCA gene selection algorithm has good classification accuracy and model stability.