Microarray gene retrieval system based on LFDA and SVM

The DNA microarray technology enables the biologists to observe the expressions of multiple thousands of genes in parallel fashion. However, processing and gaining knowledge from the voluminous microarray gene data is serious issue. It is necessary for the biologists to retrieve the required data in a reasonable time. In order to address this issue, this work presents a gene retrieval system, which is based on feature dimensionality minimization and classification of the microarray gene data. The feature dimensionality minimization is achieved by Local Fisher Discriminant Analysis (LFDA), which inherits the merits of both Fisher Discriminant Analysis (FDA) and Locality Preserving Projection (LPP). Support Vector Machine (SVM) is employed as the classifier to classify between the genes. The LFDA is chosen for reducing the dimensionality of the features, owing to its better performance on multimodal data. The SVM is trained with the feature dimensionality reduced microarray gene data, which improves the efficiency and overthrows the computational complexity. The performance of the proposed approach is compared with the LPP and FDA. Additionally, the performance of SVM is compared with the k-Nearest Neighbour (k-NN) classifier. The combination of LFDA and SVM serves better in terms of accuracy, sensitivity and specificity.