Robust Hierarchical Clustering for Metabolomics Data Analysis in presence of Cell-wise and Case-wise outliers

Abstract

Metabolomics is the sophisticated and high-throughput technology based on the entire set of metabolites which is known as the connector between genotypes and phenotypes. So, metabolomics clustering is important for metabolomics analysis. Hierarchical clustering algorithms are widely used successful unsupervised technique for analyzing Metabolomics data. Metabolomics data are collected using high-throughput technology that provides high dimensional data matrix which may contaminated by cell-wise and case-wise outliers. Traditional hierarchical clustering algorithms are highly sensitive to outliers and misleading clustering results in presence of those outliers. In this paper, an attempt is made to robustify hierarchical clustering algorithm using covariance matrix of Two Stage Generalized S-estimator (TSGS). Simulation study clearly indicates that the proposed method significantly improves the performance over the traditional hierarchical clustering approach in presence of those outliers; and almost same in absence of outliers.