A Partial least squares-based regression approach for analysis of frontotemporal dementia gene markers in human brain gene microarray data

Abstract

Conventional procedures for preliminary diagnosis of Alzheimer's disease (AD) are invasive and painful. It is important to devise noninvasive biomarker which can provide conclusive diagnosis of early onset of AD and mild cognitive impairment (MCI). Recent attention has been drawn recently to gene microarray analysis for understanding disease onset and progression. In this paper, we extend our previous work to develop a new large-scale partial least squares based multivariate regression approach for the identification of putative interacting partners of gene markers for high-throughput gene microarray and other related data. Preliminary analysis of the interacting gene partners of a marker gene of frontotemporal dementia show that the identified genes are significantly enriched in innate immune and inflammatory response processes, which align well with the nature of the disease. These suggest that the proposed approach may serve as a valuable tool for inferring putative gene interacting partners in biological studies involving gene microarray data and other related datasets.