MDAS: An integrated system for metabonomic data analysis

Juan Liu,Bo Li,Jiang-Hui Xiong

DOI: https://doi.org/10.1007/s12539-008-0008-3

2009-01-01

Abstract:Metabonomics, the latest ‘omics’ research field, shows great promise as a tool in biomarker discovery, drug efficacy and toxicity analysis, disease diagnosis and prognosis. One of the major challenges now facing researchers is how to process this data to yield useful information about a biological system, e.g., the mechanism of diseases. Traditional methods employed in metabonomic data analysis use multivariate analysis methods developed independently in chemometrics research. Additionally, with the development of machine learning approaches, some methods such as SVMs also show promise for use in metabonomic data analysis. Aside from the application of general multivariate analysis and machine learning methods to this problem, there is also a need for an integrated tool customized for metabonomic data analysis which can be easily used by biologists to reveal interesting patterns in metabonomic data. In this paper, we present a novel software tool MDAS (Metabonomic Data Analysis System) for metabonomic data analysis which integrates traditional chemometrics methods and newly introduced machine learning approaches. MDAS contains a suite of functional models for metabonomic data analysis and optimizes the flow of data analysis. Several file formats can be accepted as input. The input data can be optionally preprocessed and can then be processed with operations such as feature analysis and dimensionality reduction. The data with reduced dimensionalities can be used for training or testing through machine learning models. The system supplies proper visualization for data preprocessing, feature analysis, and classification which can be a powerful function for users to extract knowledge from the data. MDAS is an integrated platform for metabonomic data analysis, which transforms a complex analysis procedure into a more formalized and simplified one. The software package can be obtained from the authors.