Software Tools and Algorithms for Biological Systems
“Software Tools and Algorithms for Biological Systems" is composed of a collection of papers received in response to an announcement that was widely distributed to academicians and practitioners in the broad area of computational biology and software tool
- PDF / 339,420 Bytes
- 9 Pages / 439 x 666 pts Page_size
- 98 Downloads / 197 Views
Modelling Short Time Series in Metabolomics: A Functional Data Analysis Approach Giovanni Montana, Maurice Berk, and Tim Ebbels
Abstract Metabolomics is the study of the complement of small molecule metabolites in cells, biofluids and tissues. Many metabolomic experiments are designed to compare changes observed over time under two or more experimental conditions (e.g. a control and drug-treated group), thus producing time course data. Models from traditional time series analysis are often unsuitable because, by design, only very few time points are available and there are a high number of missing values. We propose a functional data analysis approach for modelling short time series arising in metabolomic studies which overcomes these obstacles. Our model assumes that each observed time series is a smooth random curve, and we propose a statistical approach for inferring this curve from repeated measurements taken on the experimental units. A test statistic for detecting differences between temporal profiles associated with two experimental conditions is then presented. The methodology has been applied to NMR spectroscopy data collected in a pre-clinical toxicology study.
1 Introduction Metabolomics (also metabonomics or metabolic profiling) is the study of the complement of small molecule metabolites in cells, biofluids and tissues [7]. Along with transcriptomics and proteomics, it is an important component of systems biology approaches which often use several such “omics” technologies to report the state of an organism at multiple biomolecular levels. Metabolomics experiments usually use sophisticated analytical techniques such as nuclear magnetic resonance (NMR) spectroscopy or mass spectrometry (MS) to assay metabolite levels, and these instruments produce large volumes of highly complex but information-rich data. An important component of all omics studies, particularly for metabolomics,
G. Montana () Mathematics, Imperial College, London, UK e-mail: [email protected]
H.R. Arabnia and Q.-N. Tran (eds.), Software Tools and Algorithms for Biological Systems, Advances in Experimental Medicine and Biology 696, c Springer Science+Business Media, LLC 2011 DOI 10.1007/978-1-4419-7046-6 31,
307
308
G. Montana et al.
is the statistical and bioinformatic methods that are used to process and model the data. In recent years, there has been much interest in developing algorithms for both low level processing (e.g. peak detection, baseline correction) and higher level modelling such as classification and regression [2]. Time is an extremely important variable in biological experiments, and many metabolomics studies produce time course data. A common experimental design consists of collecting repeated measurements in two or more experimental groups, for instance a control group and a drug-treated group; the scientific interest lies in detecting metabolites whose temporal response appears to be different, compared to the controls, due to treatment. Other related tasks include classification of individ
Data Loading...
